diff --git a/codegen/templates/RegisterCodegenUnboxedKernels.cpp b/codegen/templates/RegisterCodegenUnboxedKernels.cpp
index 3076cde1a99..180baf9b2a9 100644
--- a/codegen/templates/RegisterCodegenUnboxedKernels.cpp
+++ b/codegen/templates/RegisterCodegenUnboxedKernels.cpp
@@ -22,8 +22,8 @@
 // JIT op registry instead of c10 dispatcher. JIT op registry only takes boxed
 // kernels, so we are calling unboxing functions in UnboxingFunctions.h to cast
 // arguments into C++ types (instead of IValue) and delegate to unboxed kernels.
-using KernelSpan =
-    ::executorch::runtime::Span<const ::executorch::runtime::Kernel>;
+using KernelSpan = ::executorch::runtime::Span<
+    const ::executorch::ET_RUNTIME_NAMESPACE::Kernel>;
 namespace torch {
 namespace executor {
 namespace function {
diff --git a/devtools/bundled_program/bundled_program.cpp b/devtools/bundled_program/bundled_program.cpp
index f12262f7dd0..4e5beb84615 100644
--- a/devtools/bundled_program/bundled_program.cpp
+++ b/devtools/bundled_program/bundled_program.cpp
@@ -27,9 +27,9 @@ using executorch::aten::ArrayRef;
 using executorch::aten::Half;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
+using ::executorch::ET_RUNTIME_NAMESPACE::Method;
 using ::executorch::runtime::Error;
 using ::executorch::runtime::EValue;
-using ::executorch::runtime::Method;
 using ::executorch::runtime::Result;
 
 namespace executorch {
@@ -332,8 +332,9 @@ ET_NODISCARD Error load_bundled_input(
         static_cast<uint32_t>(status));
   }
 
-  ::executorch::runtime::internal::event_tracer_set_bundled_input_index(
-      method.get_event_tracer(), testset_idx);
+  ::executorch::ET_RUNTIME_NAMESPACE::internal::
+      event_tracer_set_bundled_input_index(
+          method.get_event_tracer(), testset_idx);
 
   return Error::Ok;
 }
diff --git a/devtools/bundled_program/bundled_program.h b/devtools/bundled_program/bundled_program.h
index 884ca6f21bc..cedb2448eca 100644
--- a/devtools/bundled_program/bundled_program.h
+++ b/devtools/bundled_program/bundled_program.h
@@ -18,7 +18,7 @@ namespace bundled_program {
  * An opaque pointer to a serialized bundled program.
  */
 using SerializedBundledProgram = const void;
-
+using ::executorch::ET_RUNTIME_NAMESPACE::Method;
 /**
  * Load testset_idx-th bundled input of method_idx-th Method test in
  * bundled_program_ptr to given Method.
@@ -31,7 +31,7 @@ using SerializedBundledProgram = const void;
  * execution.
  */
 ET_NODISCARD ::executorch::runtime::Error load_bundled_input(
-    ::executorch::runtime::Method& method,
+    Method& method,
     SerializedBundledProgram* bundled_program_ptr,
     size_t testset_idx);
 
@@ -49,7 +49,7 @@ ET_NODISCARD ::executorch::runtime::Error load_bundled_input(
  * execution.
  */
 ET_NODISCARD ::executorch::runtime::Error verify_method_outputs(
-    ::executorch::runtime::Method& method,
+    Method& method,
     SerializedBundledProgram* bundled_program_ptr,
     size_t testset_idx,
     double rtol = 1e-5,
@@ -106,7 +106,7 @@ using serialized_bundled_program =
     ::executorch::bundled_program::SerializedBundledProgram;
 
 ET_NODISCARD inline ::executorch::runtime::Error LoadBundledInput(
-    ::executorch::runtime::Method& method,
+    Method& method,
     serialized_bundled_program* bundled_program_ptr,
     size_t testset_idx) {
   return ::executorch::bundled_program::load_bundled_input(
@@ -115,7 +115,7 @@ ET_NODISCARD inline ::executorch::runtime::Error LoadBundledInput(
 
 ET_NODISCARD inline ::executorch::runtime::Error
 VerifyResultWithBundledExpectedOutput(
-    ::executorch::runtime::Method& method,
+    Method& method,
     serialized_bundled_program* bundled_program_ptr,
     size_t testset_idx,
     double rtol = 1e-5,
diff --git a/examples/portable/custom_ops/custom_ops_2_out.cpp b/examples/portable/custom_ops/custom_ops_2_out.cpp
index 138a8eeed89..2fb50e521c1 100644
--- a/examples/portable/custom_ops/custom_ops_2_out.cpp
+++ b/examples/portable/custom_ops/custom_ops_2_out.cpp
@@ -13,7 +13,7 @@ namespace native {
 
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 
 namespace {
 void check_preconditions(const Tensor& in, Tensor& out) {
diff --git a/extension/flat_tensor/flat_tensor_data_map.cpp b/extension/flat_tensor/flat_tensor_data_map.cpp
index bf54ae014b5..8aa0af13928 100644
--- a/extension/flat_tensor/flat_tensor_data_map.cpp
+++ b/extension/flat_tensor/flat_tensor_data_map.cpp
@@ -25,8 +25,8 @@ using executorch::runtime::Result;
 using executorch::runtime::Span;
 
 using executorch::aten::ScalarType;
+using executorch::ET_RUNTIME_NAMESPACE::TensorLayout;
 using executorch::runtime::DataLoader;
-using executorch::runtime::TensorLayout;
 
 namespace executorch {
 namespace extension {
diff --git a/extension/flat_tensor/flat_tensor_data_map.h b/extension/flat_tensor/flat_tensor_data_map.h
index 972a5fa9c55..0e7aee8ffc8 100644
--- a/extension/flat_tensor/flat_tensor_data_map.h
+++ b/extension/flat_tensor/flat_tensor_data_map.h
@@ -32,7 +32,8 @@ namespace extension {
 /**
  * A NamedDataMap implementation for FlatTensor-serialized data.
  */
-class FlatTensorDataMap final : public executorch::runtime::NamedDataMap {
+class FlatTensorDataMap final
+    : public executorch::ET_RUNTIME_NAMESPACE::NamedDataMap {
  public:
   /**
    * Creates a new DataMap that wraps FlatTensor data.
@@ -51,7 +52,8 @@ class FlatTensorDataMap final : public executorch::runtime::NamedDataMap {
    * @return Error::NotFound if the key is not present.
    */
   ET_NODISCARD
-  executorch::runtime::Result<const executorch::runtime::TensorLayout>
+  executorch::runtime::Result<
+      const executorch::ET_RUNTIME_NAMESPACE::TensorLayout>
   get_metadata(const char* key) const override;
 
   /**
diff --git a/extension/flat_tensor/targets.bzl b/extension/flat_tensor/targets.bzl
index 0d49995aa6e..4ac515b7bf0 100644
--- a/extension/flat_tensor/targets.bzl
+++ b/extension/flat_tensor/targets.bzl
@@ -1,24 +1,26 @@
 load("@fbsource//xplat/executorch/build:runtime_wrapper.bzl", "runtime")
 
 def define_common_targets():
-    runtime.cxx_library(
-        name = "flat_tensor_data_map",
-        srcs = [
-            "flat_tensor_data_map.cpp",
-        ],
-        exported_headers = ["flat_tensor_data_map.h"],
-        deps = [
-            "//executorch/runtime/core:core",
-            "//executorch/runtime/core:evalue",
-            "//executorch/runtime/core:named_data_map",
-            "//executorch/runtime/core/exec_aten:lib",
-            "//executorch/runtime/core/exec_aten/util:tensor_util",
-        ],
-        exported_deps = [
-            "//executorch/extension/flat_tensor/serialize:flat_tensor_header",
-            "//executorch/extension/flat_tensor/serialize:generated_headers",
-        ],
-        visibility = [
-            "//executorch/...",
-        ],
-    )
+    for aten_mode in [True, False]:
+        aten_suffix = "_aten" if aten_mode else ""
+        runtime.cxx_library(
+            name = "flat_tensor_data_map" + aten_suffix,
+            srcs = [
+                "flat_tensor_data_map.cpp",
+            ],
+            exported_headers = ["flat_tensor_data_map.h"],
+            deps = [
+                "//executorch/runtime/core:core",
+                "//executorch/runtime/core:evalue",
+                "//executorch/runtime/core:named_data_map" + aten_suffix,
+                "//executorch/runtime/core/exec_aten:lib" + aten_suffix,
+                "//executorch/runtime/core/exec_aten/util:tensor_util",
+            ],
+            exported_deps = [
+                "//executorch/extension/flat_tensor/serialize:flat_tensor_header",
+                "//executorch/extension/flat_tensor/serialize:generated_headers",
+            ],
+            visibility = [
+                "//executorch/...",
+            ],
+        )
diff --git a/extension/module/module.cpp b/extension/module/module.cpp
index 400a2c45049..6c534b8d560 100644
--- a/extension/module/module.cpp
+++ b/extension/module/module.cpp
@@ -37,6 +37,9 @@
 namespace executorch {
 namespace extension {
 
+using ET_RUNTIME_NAMESPACE::MethodMeta;
+using ET_RUNTIME_NAMESPACE::Program;
+
 namespace {
 runtime::Result<std::unique_ptr<runtime::DataLoader>> load_file(
     const std::string& file_path,
@@ -113,7 +116,7 @@ Module::Module(
 }
 
 Module::Module(
-    std::shared_ptr<runtime::Program> program,
+    std::shared_ptr<Program> program,
     std::unique_ptr<runtime::MemoryAllocator> memory_allocator,
     std::unique_ptr<runtime::MemoryAllocator> temp_allocator,
     std::unique_ptr<runtime::EventTracer> event_tracer,
@@ -131,7 +134,7 @@ Module::Module(
   runtime::runtime_init();
 }
 
-runtime::Error Module::load(const runtime::Program::Verification verification) {
+runtime::Error Module::load(const Program::Verification verification) {
   if (!is_loaded()) {
     // Load the program
     if (!data_loader_) {
@@ -156,10 +159,10 @@ runtime::Error Module::load(const runtime::Program::Verification verification) {
     }
     // else: either the map itself was provided or we have no data map, either
     // way no work to do.
-    auto program = ET_UNWRAP_UNIQUE(
-        runtime::Program::load(data_loader_.get(), verification));
-    program_ = std::shared_ptr<runtime::Program>(
-        program.release(), [](runtime::Program* pointer) { delete pointer; });
+    auto program =
+        ET_UNWRAP_UNIQUE(Program::load(data_loader_.get(), verification));
+    program_ = std::shared_ptr<Program>(
+        program.release(), [](Program* pointer) { delete pointer; });
   }
   return runtime::Error::Ok;
 }
@@ -224,7 +227,7 @@ runtime::Error Module::load_method(
   return runtime::Error::Ok;
 }
 
-runtime::Result<runtime::MethodMeta> Module::method_meta(
+runtime::Result<MethodMeta> Module::method_meta(
     const std::string& method_name) {
   ET_CHECK_OK_OR_RETURN_ERROR(load_method(method_name));
   return methods_.at(method_name).method->method_meta();
diff --git a/extension/module/module.h b/extension/module/module.h
index 45d2cc1d14b..73c7328ee0a 100644
--- a/extension/module/module.h
+++ b/extension/module/module.h
@@ -19,6 +19,11 @@
 namespace executorch {
 namespace extension {
 
+using ET_RUNTIME_NAMESPACE::Method;
+using ET_RUNTIME_NAMESPACE::MethodMeta;
+using ET_RUNTIME_NAMESPACE::NamedDataMap;
+using ET_RUNTIME_NAMESPACE::Program;
+
 /**
  * A facade class for loading programs and executing methods within them.
  */
@@ -95,7 +100,7 @@ class Module {
    * @param[in] data_map_loader A DataLoader used for loading external weights.
    */
   explicit Module(
-      std::shared_ptr<runtime::Program> program,
+      std::shared_ptr<Program> program,
       std::unique_ptr<runtime::MemoryAllocator> memory_allocator = nullptr,
       std::unique_ptr<runtime::MemoryAllocator> temp_allocator = nullptr,
       std::unique_ptr<runtime::EventTracer> event_tracer = nullptr,
@@ -116,8 +121,8 @@ class Module {
    */
   ET_NODISCARD
   runtime::Error load(
-      const runtime::Program::Verification verification =
-          runtime::Program::Verification::Minimal);
+      const Program::Verification verification =
+          Program::Verification::Minimal);
 
   /**
    * Checks if the program is loaded.
@@ -134,7 +139,7 @@ class Module {
    *
    * @returns Shared pointer to the program or nullptr if it's not yet loaded.
    */
-  inline std::shared_ptr<runtime::Program> program() const {
+  inline std::shared_ptr<Program> program() const {
     return program_;
   }
 
@@ -224,8 +229,7 @@ class Module {
    * @returns A method metadata, or an error if the program or method failed to
    * load.
    */
-  runtime::Result<runtime::MethodMeta> method_meta(
-      const std::string& method_name);
+  runtime::Result<MethodMeta> method_meta(const std::string& method_name);
 
   /**
    * Execute a specific method with the given input values and retrieve the
@@ -473,20 +477,20 @@ class Module {
     std::vector<runtime::Span<uint8_t>> planned_spans;
     std::unique_ptr<runtime::HierarchicalAllocator> planned_memory;
     std::unique_ptr<runtime::MemoryManager> memory_manager;
-    std::unique_ptr<runtime::Method> method;
+    std::unique_ptr<Method> method;
     std::vector<runtime::EValue> inputs;
   };
 
   std::string file_path_;
   std::string data_map_path_;
   LoadMode load_mode_{LoadMode::MmapUseMlock};
-  std::shared_ptr<runtime::Program> program_;
+  std::shared_ptr<Program> program_;
   std::unique_ptr<runtime::DataLoader> data_loader_;
   std::unique_ptr<runtime::MemoryAllocator> memory_allocator_;
   std::unique_ptr<runtime::MemoryAllocator> temp_allocator_;
   std::unique_ptr<runtime::EventTracer> event_tracer_;
   std::unique_ptr<runtime::DataLoader> data_map_loader_;
-  std::unique_ptr<runtime::NamedDataMap> data_map_;
+  std::unique_ptr<NamedDataMap> data_map_;
 
  protected:
   std::unordered_map<std::string, MethodHolder> methods_;
diff --git a/extension/module/targets.bzl b/extension/module/targets.bzl
index 09a610a1fca..d8019ce9c4e 100644
--- a/extension/module/targets.bzl
+++ b/extension/module/targets.bzl
@@ -25,7 +25,7 @@ def define_common_targets():
                 "//executorch/extension/memory_allocator:malloc_memory_allocator",
                 "//executorch/extension/data_loader:file_data_loader",
                 "//executorch/extension/data_loader:mmap_data_loader",
-                "//executorch/extension/flat_tensor:flat_tensor_data_map",
+                "//executorch/extension/flat_tensor:flat_tensor_data_map" + aten_suffix,
             ],
             exported_deps = [
                 "//executorch/runtime/executor:program" + aten_suffix,
diff --git a/extension/pybindings/pybindings.cpp b/extension/pybindings/pybindings.cpp
index a998e591f30..766288d7e47 100644
--- a/extension/pybindings/pybindings.cpp
+++ b/extension/pybindings/pybindings.cpp
@@ -85,26 +85,26 @@ void et_pal_emit_log_message(
 
 namespace py = pybind11;
 using executorch::bundled_program::verify_method_outputs;
+using ::executorch::ET_RUNTIME_NAMESPACE::BackendInterface;
+using ::executorch::ET_RUNTIME_NAMESPACE::get_backend_class;
+using ::executorch::ET_RUNTIME_NAMESPACE::get_backend_name;
+using ::executorch::ET_RUNTIME_NAMESPACE::get_num_registered_backends;
+using ::executorch::ET_RUNTIME_NAMESPACE::get_registered_kernels;
+using ::executorch::ET_RUNTIME_NAMESPACE::Kernel;
+using ::executorch::ET_RUNTIME_NAMESPACE::Method;
+using ::executorch::ET_RUNTIME_NAMESPACE::Program;
 using ::executorch::extension::BufferDataLoader;
 using ::executorch::extension::MallocMemoryAllocator;
 using ::executorch::extension::MmapDataLoader;
 using ::executorch::runtime::ArrayRef;
-using ::executorch::runtime::BackendInterface;
 using ::executorch::runtime::DataLoader;
 using ::executorch::runtime::Error;
 using ::executorch::runtime::EValue;
 using ::executorch::runtime::EventTracerDebugLogLevel;
-using ::executorch::runtime::get_backend_class;
-using ::executorch::runtime::get_backend_name;
-using ::executorch::runtime::get_num_registered_backends;
-using ::executorch::runtime::get_registered_kernels;
 using ::executorch::runtime::HierarchicalAllocator;
-using ::executorch::runtime::Kernel;
 using ::executorch::runtime::MemoryAllocator;
 using ::executorch::runtime::MemoryManager;
-using ::executorch::runtime::Method;
 using ::executorch::runtime::prof_result_t;
-using ::executorch::runtime::Program;
 using ::executorch::runtime::Result;
 using ::executorch::runtime::Span;
 using ::executorch::runtime::Tag;
diff --git a/extension/runner_util/inputs.cpp b/extension/runner_util/inputs.cpp
index 11cd176b5d1..842ba25532f 100644
--- a/extension/runner_util/inputs.cpp
+++ b/extension/runner_util/inputs.cpp
@@ -12,12 +12,12 @@
 #include <executorch/runtime/executor/method_meta.h>
 #include <executorch/runtime/platform/log.h>
 
+using executorch::ET_RUNTIME_NAMESPACE::Method;
+using executorch::ET_RUNTIME_NAMESPACE::MethodMeta;
+using executorch::ET_RUNTIME_NAMESPACE::TensorInfo;
 using executorch::runtime::Error;
-using executorch::runtime::Method;
-using executorch::runtime::MethodMeta;
 using executorch::runtime::Result;
 using executorch::runtime::Tag;
-using executorch::runtime::TensorInfo;
 
 namespace executorch {
 namespace extension {
diff --git a/extension/runner_util/inputs.h b/extension/runner_util/inputs.h
index 73722c0d7bf..214b76d67e3 100644
--- a/extension/runner_util/inputs.h
+++ b/extension/runner_util/inputs.h
@@ -15,6 +15,9 @@
 namespace executorch {
 namespace extension {
 
+using ::executorch::ET_RUNTIME_NAMESPACE::Method;
+using ::executorch::ET_RUNTIME_NAMESPACE::TensorInfo;
+
 /**
  * RAII helper that frees a set of buffers when destroyed. Movable.
  */
@@ -80,7 +83,7 @@ struct PrepareInputTensorsOptions {
  * @returns An error on failure.
  */
 executorch::runtime::Result<BufferCleanup> prepare_input_tensors(
-    executorch::runtime::Method& method,
+    Method& method,
     PrepareInputTensorsOptions options = {});
 
 namespace internal {
@@ -89,8 +92,8 @@ namespace internal {
  * fills it with ones, and sets the input at `input_index`.
  */
 executorch::runtime::Error fill_and_set_input(
-    executorch::runtime::Method& method,
-    executorch::runtime::TensorInfo& tensor_meta,
+    Method& method,
+    TensorInfo& tensor_meta,
     size_t input_index,
     void* data_ptr);
 } // namespace internal
diff --git a/extension/runner_util/inputs_aten.cpp b/extension/runner_util/inputs_aten.cpp
index 83d12dac42d..b89562a2f69 100644
--- a/extension/runner_util/inputs_aten.cpp
+++ b/extension/runner_util/inputs_aten.cpp
@@ -15,8 +15,8 @@
 #include <executorch/runtime/executor/method_meta.h>
 
 using executorch::runtime::Error;
-using executorch::runtime::Method;
-using executorch::runtime::TensorInfo;
+using executorch::runtime::aten::Method;
+using executorch::runtime::aten::TensorInfo;
 
 namespace executorch {
 namespace extension {
diff --git a/extension/tensor/tensor_ptr.cpp b/extension/tensor/tensor_ptr.cpp
index c1742fc599b..8a35e83a526 100644
--- a/extension/tensor/tensor_ptr.cpp
+++ b/extension/tensor/tensor_ptr.cpp
@@ -188,7 +188,7 @@ TensorPtr clone_tensor_ptr(const executorch::aten::Tensor& tensor) {
 runtime::Error resize_tensor_ptr(
     TensorPtr& tensor,
     const std::vector<executorch::aten::SizesType>& sizes) {
-  return runtime::resize_tensor(
+  return ET_RUNTIME_NAMESPACE::resize_tensor(
       *tensor,
       executorch::aten::ArrayRef<executorch::aten::SizesType>(
           sizes.data(), sizes.size()));
diff --git a/kernels/aten/cpu/op__empty_dim_order.cpp b/kernels/aten/cpu/op__empty_dim_order.cpp
index e75963a9c4e..654b29c778d 100644
--- a/kernels/aten/cpu/op__empty_dim_order.cpp
+++ b/kernels/aten/cpu/op__empty_dim_order.cpp
@@ -102,7 +102,7 @@ Tensor& _empty_dim_order_out(
     IntArrayRef size,
     OptionalIntArrayRef dim_order,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext ctx{};
+  KernelRuntimeContext ctx{};
   return _empty_dim_order_out(ctx, size, dim_order, out);
 }
 
diff --git a/kernels/aten/cpu/op__to_dim_order_copy.cpp b/kernels/aten/cpu/op__to_dim_order_copy.cpp
index 10793d24db5..a8216c9a8e9 100644
--- a/kernels/aten/cpu/op__to_dim_order_copy.cpp
+++ b/kernels/aten/cpu/op__to_dim_order_copy.cpp
@@ -116,7 +116,7 @@ Tensor& _to_dim_order_copy_out(
     bool non_blocking,
     OptionalArrayRef<int64_t> dim_order,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext ctx{};
+  KernelRuntimeContext ctx{};
   return _to_dim_order_copy_out(ctx, self, non_blocking, dim_order, out);
 }
 
diff --git a/kernels/portable/cpu/op__to_dim_order_copy.cpp b/kernels/portable/cpu/op__to_dim_order_copy.cpp
index 40ce86e8fdc..70fc3507f05 100644
--- a/kernels/portable/cpu/op__to_dim_order_copy.cpp
+++ b/kernels/portable/cpu/op__to_dim_order_copy.cpp
@@ -125,7 +125,7 @@ Tensor& _to_dim_order_copy_out(
     bool non_blocking,
     OptionalArrayRef<int64_t> dim_order,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return _to_dim_order_copy_out(context, self, non_blocking, dim_order, out);
 }
 
diff --git a/kernels/portable/test/op_gelu_test.cpp b/kernels/portable/test/op_gelu_test.cpp
index 19e757b4bd0..2e5cad55c35 100644
--- a/kernels/portable/test/op_gelu_test.cpp
+++ b/kernels/portable/test/op_gelu_test.cpp
@@ -25,7 +25,7 @@ using torch::executor::testing::TensorFactory;
 // executorch/kernels/test/op_gelu_test.cpp instead.
 
 Tensor& op_gelu_out(const Tensor& self, string_view approximate, Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::native::gelu_out(context, self, approximate, out);
 }
 
diff --git a/kernels/prim_ops/register_prim_ops.cpp b/kernels/prim_ops/register_prim_ops.cpp
index 1d197b63584..62aead8978f 100644
--- a/kernels/prim_ops/register_prim_ops.cpp
+++ b/kernels/prim_ops/register_prim_ops.cpp
@@ -381,14 +381,13 @@ static Kernel prim_ops[] = {
 
 };
 
-executorch::runtime::Span<const executorch::runtime::Kernel> kernel_span(
-    prim_ops,
-    prim_ops + sizeof(prim_ops) / sizeof(Kernel));
+executorch::runtime::Span<const executorch::ET_RUNTIME_NAMESPACE::Kernel>
+    kernel_span(prim_ops, prim_ops + sizeof(prim_ops) / sizeof(Kernel));
 
 // Return value not used. Keep the static variable assignment to register
 // operators in static initialization time.
 auto success_with_kernel_reg =
-    executorch::runtime::register_kernels(kernel_span);
+    executorch::ET_RUNTIME_NAMESPACE::register_kernels(kernel_span);
 
 } // namespace
 } // namespace function
diff --git a/kernels/prim_ops/targets.bzl b/kernels/prim_ops/targets.bzl
index c1af21a7e73..d2cff10a194 100644
--- a/kernels/prim_ops/targets.bzl
+++ b/kernels/prim_ops/targets.bzl
@@ -56,7 +56,7 @@ def define_common_targets():
                 ":et_copy_index" + aten_suffix,
                 ":et_view" + aten_suffix,
                 "//executorch/runtime/core:evalue" + aten_suffix,
-                "//executorch/runtime/kernel:operator_registry",
+                "//executorch/runtime/kernel:operator_registry" + aten_suffix,
                 "//executorch/runtime/kernel:kernel_includes" + aten_suffix,
             ],
         )
diff --git a/kernels/quantized/cpu/op_quantize.cpp b/kernels/quantized/cpu/op_quantize.cpp
index 5079109683f..632bddd58c4 100644
--- a/kernels/quantized/cpu/op_quantize.cpp
+++ b/kernels/quantized/cpu/op_quantize.cpp
@@ -22,6 +22,7 @@ namespace native {
 using Tensor = executorch::aten::Tensor;
 using Scalar = executorch::aten::Scalar;
 using ScalarType = executorch::aten::ScalarType;
+using KernelRuntimeContext = torch::executor::KernelRuntimeContext;
 
 namespace {
 
@@ -214,7 +215,7 @@ Tensor& quantize_per_tensor_tensor_args_out(
     int64_t quant_max,
     ScalarType dtype,
     Tensor& out) {
-  auto context = executorch::runtime::KernelRuntimeContext();
+  auto context = KernelRuntimeContext();
   auto& res = quantize_per_tensor_tensor_args_out(
       context, input, scale, zero_point, quant_min, quant_max, dtype, out);
   ET_CHECK(context.failure_state() == Error::Ok);
diff --git a/kernels/quantized/test/op_add_test.cpp b/kernels/quantized/test/op_add_test.cpp
index 17dd1cfb3fc..3f258827973 100644
--- a/kernels/quantized/test/op_add_test.cpp
+++ b/kernels/quantized/test/op_add_test.cpp
@@ -24,7 +24,7 @@ using executorch::aten::optional;
 using executorch::aten::Scalar;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::native::add_out;
 using torch::executor::native::dequantize_per_tensor_out;
 using torch::executor::native::quantize_per_tensor_out;
diff --git a/kernels/quantized/test/op_embedding2b_test.cpp b/kernels/quantized/test/op_embedding2b_test.cpp
index bf48fa4227b..a350b77ec0d 100644
--- a/kernels/quantized/test/op_embedding2b_test.cpp
+++ b/kernels/quantized/test/op_embedding2b_test.cpp
@@ -21,7 +21,7 @@ using executorch::aten::ArrayRef;
 using executorch::aten::optional;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::native::quantized_embedding_2bit_out;
 
 using torch::executor::testing::TensorFactory;
diff --git a/kernels/quantized/test/op_embedding4b_test.cpp b/kernels/quantized/test/op_embedding4b_test.cpp
index 9f205be80e3..6ab10376b88 100644
--- a/kernels/quantized/test/op_embedding4b_test.cpp
+++ b/kernels/quantized/test/op_embedding4b_test.cpp
@@ -21,7 +21,7 @@ using executorch::aten::ArrayRef;
 using executorch::aten::optional;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::native::quantized_embedding_4bit_out;
 
 using torch::executor::testing::TensorFactory;
diff --git a/kernels/quantized/test/op_embedding_test.cpp b/kernels/quantized/test/op_embedding_test.cpp
index 252aca41314..6c949bd6e69 100644
--- a/kernels/quantized/test/op_embedding_test.cpp
+++ b/kernels/quantized/test/op_embedding_test.cpp
@@ -24,7 +24,7 @@ using executorch::aten::optional;
 using executorch::aten::Scalar;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::native::dequantize_per_tensor_out;
 using torch::executor::native::embedding_out;
 using torch::executor::native::quantize_per_tensor_out;
diff --git a/kernels/quantized/test/op_mixed_linear_test.cpp b/kernels/quantized/test/op_mixed_linear_test.cpp
index 6b86b199f60..833fc766ffd 100644
--- a/kernels/quantized/test/op_mixed_linear_test.cpp
+++ b/kernels/quantized/test/op_mixed_linear_test.cpp
@@ -20,7 +20,7 @@ using namespace ::testing;
 using executorch::aten::optional;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::native::quantized_mixed_linear_out;
 using torch::executor::testing::TensorFactory;
 
diff --git a/kernels/quantized/test/op_mixed_mm_test.cpp b/kernels/quantized/test/op_mixed_mm_test.cpp
index e20ac96d610..4d81089fa91 100644
--- a/kernels/quantized/test/op_mixed_mm_test.cpp
+++ b/kernels/quantized/test/op_mixed_mm_test.cpp
@@ -20,7 +20,7 @@ using namespace ::testing;
 using executorch::aten::optional;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::native::quantized_mixed_mm_out;
 using torch::executor::testing::TensorFactory;
 
diff --git a/kernels/test/TestUtil.h b/kernels/test/TestUtil.h
index aa220f5bfd5..7ec20c11bef 100644
--- a/kernels/test/TestUtil.h
+++ b/kernels/test/TestUtil.h
@@ -116,6 +116,6 @@ class OperatorTest : public ::testing::Test {
   }
 
  protected:
-  executorch::runtime::KernelRuntimeContext context_;
+  ::torch::executor::KernelRuntimeContext context_;
   bool expect_failure_;
 };
diff --git a/kernels/test/custom_kernel_example/op_relu.cpp b/kernels/test/custom_kernel_example/op_relu.cpp
index 2cc3eefe0a8..074ebe6b900 100644
--- a/kernels/test/custom_kernel_example/op_relu.cpp
+++ b/kernels/test/custom_kernel_example/op_relu.cpp
@@ -17,8 +17,8 @@ namespace native {
 
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using executorch::runtime::Error;
-using executorch::runtime::KernelRuntimeContext;
 using executorch::runtime::resize_tensor;
 using executorch::runtime::tensors_have_same_shape_and_dtype;
 
diff --git a/kernels/test/op_atan2_test.cpp b/kernels/test/op_atan2_test.cpp
index e69ea0e90c8..436826e2b6d 100644
--- a/kernels/test/op_atan2_test.cpp
+++ b/kernels/test/op_atan2_test.cpp
@@ -23,7 +23,7 @@ using torch::executor::testing::SupportedFeatures;
 using torch::executor::testing::TensorFactory;
 
 Tensor& op_atan2_out(const Tensor& self, const Tensor& other, Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::atan2_outf(context, self, other, out);
 }
 
diff --git a/kernels/test/op_cdist_forward_test.cpp b/kernels/test/op_cdist_forward_test.cpp
index 32465ca439b..9ddab4c3c49 100644
--- a/kernels/test/op_cdist_forward_test.cpp
+++ b/kernels/test/op_cdist_forward_test.cpp
@@ -21,6 +21,7 @@ using executorch::aten::ArrayRef;
 using executorch::aten::optional;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::testing::TensorFactory;
 
 Tensor& op_cdist_forward_out(
@@ -29,7 +30,7 @@ Tensor& op_cdist_forward_out(
     double p,
     optional<int64_t> compute_mode,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  KernelRuntimeContext context{};
   return torch::executor::aten::_cdist_forward_outf(
       context, x1, x2, p, compute_mode, out);
 }
diff --git a/kernels/test/op_clamp_test.cpp b/kernels/test/op_clamp_test.cpp
index a1003e892e0..8a021c70303 100644
--- a/kernels/test/op_clamp_test.cpp
+++ b/kernels/test/op_clamp_test.cpp
@@ -260,7 +260,7 @@ class OpClampTensorOutTest : public OperatorTest {
       const optional<Tensor>& min,
       const optional<Tensor>& max,
       Tensor& out) {
-    executorch::runtime::KernelRuntimeContext context{};
+    executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
     return torch::executor::aten::clamp_outf(context, self, min, max, out);
   }
 };
diff --git a/kernels/test/op_diagonal_copy_test.cpp b/kernels/test/op_diagonal_copy_test.cpp
index cc0bd02e1a5..35d19bf661e 100644
--- a/kernels/test/op_diagonal_copy_test.cpp
+++ b/kernels/test/op_diagonal_copy_test.cpp
@@ -27,7 +27,7 @@ Tensor& op_diagonal_copy_out(
     int64_t dim1,
     int64_t dim2,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::diagonal_copy_outf(
       context, input, offset, dim1, dim2, out);
 }
diff --git a/kernels/test/op_flip_test.cpp b/kernels/test/op_flip_test.cpp
index f240dfd4ad3..be06e397be2 100644
--- a/kernels/test/op_flip_test.cpp
+++ b/kernels/test/op_flip_test.cpp
@@ -22,7 +22,7 @@ using executorch::aten::Tensor;
 using torch::executor::testing::TensorFactory;
 
 Tensor& op_flip_out(const Tensor& input, IntArrayRef dims, Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::flip_outf(context, input, dims, out);
 }
 
diff --git a/kernels/test/op_ge_test.cpp b/kernels/test/op_ge_test.cpp
index 4b21644a5c5..a79502b266e 100644
--- a/kernels/test/op_ge_test.cpp
+++ b/kernels/test/op_ge_test.cpp
@@ -18,7 +18,7 @@ using namespace ::testing;
 using executorch::aten::Scalar;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::testing::TensorFactory;
 
 class OpGeTensorOutTest : public OperatorTest {
diff --git a/kernels/test/op_gt_test.cpp b/kernels/test/op_gt_test.cpp
index 29a2fb0e8b8..96c0e95f950 100644
--- a/kernels/test/op_gt_test.cpp
+++ b/kernels/test/op_gt_test.cpp
@@ -18,7 +18,7 @@ using namespace ::testing;
 using executorch::aten::Scalar;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::testing::TensorFactory;
 
 class OpGtScalarOutTest : public OperatorTest {
diff --git a/kernels/test/op_le_test.cpp b/kernels/test/op_le_test.cpp
index 49ef5235d0f..bcd40d24d89 100644
--- a/kernels/test/op_le_test.cpp
+++ b/kernels/test/op_le_test.cpp
@@ -18,7 +18,7 @@ using namespace ::testing;
 using executorch::aten::Scalar;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::testing::TensorFactory;
 
 class OpLeScalarOutTest : public OperatorTest {
diff --git a/kernels/test/op_lt_test.cpp b/kernels/test/op_lt_test.cpp
index 51ccb310e4a..eee12c50521 100644
--- a/kernels/test/op_lt_test.cpp
+++ b/kernels/test/op_lt_test.cpp
@@ -18,7 +18,7 @@ using namespace ::testing;
 using executorch::aten::Scalar;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::testing::TensorFactory;
 
 class OpLtScalarOutTest : public OperatorTest {
diff --git a/kernels/test/op_maximum_test.cpp b/kernels/test/op_maximum_test.cpp
index 9c701e208eb..faa18fa56cd 100644
--- a/kernels/test/op_maximum_test.cpp
+++ b/kernels/test/op_maximum_test.cpp
@@ -21,7 +21,7 @@ using executorch::aten::Tensor;
 using torch::executor::testing::TensorFactory;
 
 Tensor& op_maximum_out(const Tensor& self, const Tensor& other, Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::maximum_outf(context, self, other, out);
 }
 
diff --git a/kernels/test/op_native_batch_norm_test.cpp b/kernels/test/op_native_batch_norm_test.cpp
index 67e46b27508..bf05a87312d 100644
--- a/kernels/test/op_native_batch_norm_test.cpp
+++ b/kernels/test/op_native_batch_norm_test.cpp
@@ -173,7 +173,7 @@ class OpNativeBatchNormLegitOutTest : public OperatorTest {
       executorch::aten::Tensor& out0,
       executorch::aten::Tensor& out1,
       executorch::aten::Tensor& out2) {
-    executorch::runtime::KernelRuntimeContext context{};
+    executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
     return torch::executor::aten::_native_batch_norm_legit_outf(
         context,
         input,
diff --git a/kernels/test/op_native_group_norm_test.cpp b/kernels/test/op_native_group_norm_test.cpp
index ea742e97231..7452350ad29 100644
--- a/kernels/test/op_native_group_norm_test.cpp
+++ b/kernels/test/op_native_group_norm_test.cpp
@@ -32,7 +32,7 @@ ::std::tuple<Tensor&, Tensor&, Tensor&> op_native_group_norm_out(
     Tensor& out0,
     Tensor& out1,
     Tensor& out2) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::native_group_norm_outf(
       context, input, weight, bias, N, C, HxW, group, eps, out0, out1, out2);
 }
diff --git a/kernels/test/op_ne_test.cpp b/kernels/test/op_ne_test.cpp
index 6cb0217ec0f..fe4e6c3621c 100644
--- a/kernels/test/op_ne_test.cpp
+++ b/kernels/test/op_ne_test.cpp
@@ -18,7 +18,7 @@ using namespace ::testing;
 using executorch::aten::Scalar;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using torch::executor::testing::TensorFactory;
 
 class OpNeTest : public OperatorTest {
diff --git a/kernels/test/op_pdist_forward_test.cpp b/kernels/test/op_pdist_forward_test.cpp
index e6c0d472517..2b28591f7fc 100644
--- a/kernels/test/op_pdist_forward_test.cpp
+++ b/kernels/test/op_pdist_forward_test.cpp
@@ -23,7 +23,7 @@ using executorch::aten::Tensor;
 using torch::executor::testing::TensorFactory;
 
 Tensor& op_pdist_forward_out(const Tensor& input, double p, Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::_pdist_forward_outf(context, input, p, out);
 }
 
diff --git a/kernels/test/op_prod_test.cpp b/kernels/test/op_prod_test.cpp
index f9cf53ded57..11a7e3fae4f 100644
--- a/kernels/test/op_prod_test.cpp
+++ b/kernels/test/op_prod_test.cpp
@@ -23,7 +23,7 @@ using torch::executor::testing::TensorFactory;
 
 Tensor&
 op_prod_out(const Tensor& self, optional<ScalarType> dtype, Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::prod_outf(context, self, dtype, out);
 }
 
@@ -33,7 +33,7 @@ Tensor& op_prod_int_out(
     bool keepdim,
     optional<ScalarType> dtype,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::prod_outf(
       context, self, dim, keepdim, dtype, out);
 }
diff --git a/kernels/test/op_reflection_pad1d_test.cpp b/kernels/test/op_reflection_pad1d_test.cpp
index 5f3b2a1c273..aebf057326a 100644
--- a/kernels/test/op_reflection_pad1d_test.cpp
+++ b/kernels/test/op_reflection_pad1d_test.cpp
@@ -25,7 +25,7 @@ Tensor& op_reflection_pad1d_out(
     const Tensor& input,
     ArrayRef<int64_t> padding,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::reflection_pad1d_outf(
       context, input, padding, out);
 }
diff --git a/kernels/test/op_reflection_pad2d_test.cpp b/kernels/test/op_reflection_pad2d_test.cpp
index 8696b5dff7b..01e0619b9f1 100644
--- a/kernels/test/op_reflection_pad2d_test.cpp
+++ b/kernels/test/op_reflection_pad2d_test.cpp
@@ -25,7 +25,7 @@ Tensor& op_reflection_pad2d_out(
     const Tensor& input,
     ArrayRef<int64_t> padding,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::reflection_pad2d_outf(
       context, input, padding, out);
 }
diff --git a/kernels/test/op_reflection_pad3d_test.cpp b/kernels/test/op_reflection_pad3d_test.cpp
index 7d5cc84c6bc..55ed906a958 100644
--- a/kernels/test/op_reflection_pad3d_test.cpp
+++ b/kernels/test/op_reflection_pad3d_test.cpp
@@ -25,7 +25,7 @@ Tensor& op_reflection_pad3d_out(
     const Tensor& input,
     ArrayRef<int64_t> padding,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::reflection_pad3d_outf(
       context, input, padding, out);
 }
diff --git a/kernels/test/op_replication_pad1d_test.cpp b/kernels/test/op_replication_pad1d_test.cpp
index 9a6d3b2285e..f8a3fc0a48b 100644
--- a/kernels/test/op_replication_pad1d_test.cpp
+++ b/kernels/test/op_replication_pad1d_test.cpp
@@ -25,7 +25,7 @@ Tensor& op_replication_pad1d_out(
     const Tensor& input,
     ArrayRef<int64_t> padding,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::replication_pad1d_outf(
       context, input, padding, out);
 }
diff --git a/kernels/test/op_replication_pad2d_test.cpp b/kernels/test/op_replication_pad2d_test.cpp
index 00bc76ac093..7f62f5c9b6e 100644
--- a/kernels/test/op_replication_pad2d_test.cpp
+++ b/kernels/test/op_replication_pad2d_test.cpp
@@ -25,7 +25,7 @@ Tensor& op_replication_pad2d_out(
     const Tensor& input,
     ArrayRef<int64_t> padding,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::replication_pad2d_outf(
       context, input, padding, out);
 }
diff --git a/kernels/test/op_replication_pad3d_test.cpp b/kernels/test/op_replication_pad3d_test.cpp
index 010870298d9..5b931fee3f9 100644
--- a/kernels/test/op_replication_pad3d_test.cpp
+++ b/kernels/test/op_replication_pad3d_test.cpp
@@ -25,7 +25,7 @@ Tensor& op_replication_pad3d_out(
     const Tensor& input,
     ArrayRef<int64_t> padding,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::replication_pad3d_outf(
       context, input, padding, out);
 }
diff --git a/kernels/test/op_roll_test.cpp b/kernels/test/op_roll_test.cpp
index fc5baaad4a7..4407e395db6 100644
--- a/kernels/test/op_roll_test.cpp
+++ b/kernels/test/op_roll_test.cpp
@@ -26,7 +26,7 @@ Tensor& op_roll_out(
     ArrayRef<int64_t> shifts,
     ArrayRef<int64_t> dims,
     Tensor& out) {
-  executorch::runtime::KernelRuntimeContext context{};
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context{};
   return torch::executor::aten::roll_outf(context, input, shifts, dims, out);
 }
 
diff --git a/kernels/test/op_topk_test.cpp b/kernels/test/op_topk_test.cpp
index 46098a81b68..bdd185daaae 100644
--- a/kernels/test/op_topk_test.cpp
+++ b/kernels/test/op_topk_test.cpp
@@ -106,7 +106,8 @@ std::tuple<Tensor&, Tensor&> op_topk_values(
     Tensor& values,
     Tensor& indices) {
   TempMemoryAllocator allocator = TempMemoryAllocator();
-  executorch::runtime::KernelRuntimeContext context(nullptr, &allocator);
+  executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext context(
+      nullptr, &allocator);
   return torch::executor::aten::topk_outf(
       context, input, k, dim, largest, sorted, values, indices);
 }
diff --git a/runtime/backend/backend_execution_context.h b/runtime/backend/backend_execution_context.h
index d2790b158ef..7ee41d8e5b1 100644
--- a/runtime/backend/backend_execution_context.h
+++ b/runtime/backend/backend_execution_context.h
@@ -12,7 +12,7 @@
 #include <executorch/runtime/core/memory_allocator.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
 /**
  * BackendExecutionContext will be used to inject run time context.
@@ -68,13 +68,13 @@ class BackendExecutionContext final {
   const char* method_name_ = nullptr;
 };
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 namespace torch {
 namespace executor {
 // TODO(T197294990): Remove these deprecated aliases once all users have moved
 // to the new `::executorch` namespaces.
-using ::executorch::runtime::BackendExecutionContext;
+using ::executorch::ET_RUNTIME_NAMESPACE::BackendExecutionContext;
 } // namespace executor
 } // namespace torch
diff --git a/runtime/backend/backend_init_context.h b/runtime/backend/backend_init_context.h
index de1661c3af0..71c5182f401 100644
--- a/runtime/backend/backend_init_context.h
+++ b/runtime/backend/backend_init_context.h
@@ -7,12 +7,12 @@
  */
 
 #pragma once
+#include <executorch/runtime/core/event_tracer.h>
 #include <executorch/runtime/core/memory_allocator.h>
 #include <executorch/runtime/core/named_data_map.h>
 
 namespace executorch {
-namespace runtime {
-
+namespace ET_RUNTIME_NAMESPACE {
 /**
  * BackendInitContext will be used to inject runtime info for to initialize
  * delegate.
@@ -70,13 +70,13 @@ class BackendInitContext final {
   const NamedDataMap* named_data_map_ = nullptr;
 };
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 namespace torch {
 namespace executor {
 // TODO(T197294990): Remove these deprecated aliases once all users have moved
 // to the new `::executorch` namespaces.
-using ::executorch::runtime::BackendInitContext;
+using ::executorch::ET_RUNTIME_NAMESPACE::BackendInitContext;
 } // namespace executor
 } // namespace torch
diff --git a/runtime/backend/interface.cpp b/runtime/backend/interface.cpp
index 4fb1eadfa87..ffeb133fbf2 100644
--- a/runtime/backend/interface.cpp
+++ b/runtime/backend/interface.cpp
@@ -9,7 +9,7 @@
 #include <executorch/runtime/backend/interface.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
 // Pure-virtual dtors still need an implementation.
 BackendInterface::~BackendInterface() {}
@@ -66,5 +66,5 @@ Result<const char*> get_backend_name(size_t index) {
   return registered_backends[index].name;
 }
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/backend/interface.h b/runtime/backend/interface.h
index 0a3c069a201..95705d48f92 100644
--- a/runtime/backend/interface.h
+++ b/runtime/backend/interface.h
@@ -22,7 +22,7 @@
 #include <executorch/runtime/platform/compiler.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
 struct SizedBuffer {
   void* buffer;
@@ -150,19 +150,20 @@ size_t get_num_registered_backends();
  */
 Result<const char*> get_backend_name(size_t index);
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 namespace torch {
 namespace executor {
 // TODO(T197294990): Remove these deprecated aliases once all users have moved
 // to the new `::executorch` namespaces.
-using ::executorch::runtime::Backend;
-using ::executorch::runtime::CompileSpec;
-using ::executorch::runtime::DelegateHandle;
-using ::executorch::runtime::get_backend_class;
-using ::executorch::runtime::register_backend;
-using ::executorch::runtime::SizedBuffer;
-using PyTorchBackendInterface = ::executorch::runtime::BackendInterface;
+using ::executorch::ET_RUNTIME_NAMESPACE::Backend;
+using ::executorch::ET_RUNTIME_NAMESPACE::CompileSpec;
+using ::executorch::ET_RUNTIME_NAMESPACE::DelegateHandle;
+using ::executorch::ET_RUNTIME_NAMESPACE::get_backend_class;
+using ::executorch::ET_RUNTIME_NAMESPACE::register_backend;
+using ::executorch::ET_RUNTIME_NAMESPACE::SizedBuffer;
+using PyTorchBackendInterface =
+    ::executorch::ET_RUNTIME_NAMESPACE::BackendInterface;
 } // namespace executor
 } // namespace torch
diff --git a/runtime/core/event_tracer_hooks.h b/runtime/core/event_tracer_hooks.h
index 40754160c41..cd74b447ca8 100644
--- a/runtime/core/event_tracer_hooks.h
+++ b/runtime/core/event_tracer_hooks.h
@@ -30,7 +30,7 @@
  */
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 namespace internal {
 
 /**
@@ -305,7 +305,7 @@ inline void event_tracer_set_bundled_input_index(
 }
 
 } // namespace internal
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 namespace torch {
@@ -313,18 +313,27 @@ namespace executor {
 namespace internal {
 // TODO(T197294990): Remove these deprecated aliases once all users have moved
 // to the new `::executorch` namespaces.
-using ::executorch::runtime::internal::event_tracer_begin_profiling_event;
-using ::executorch::runtime::internal::event_tracer_create_event_block;
-using ::executorch::runtime::internal::event_tracer_end_profiling_event;
-using ::executorch::runtime::internal::event_tracer_log_evalue;
-using ::executorch::runtime::internal::event_tracer_log_evalue_output;
-using ::executorch::runtime::internal::event_tracer_set_bundled_input_index;
-using ::executorch::runtime::internal::event_tracer_track_allocation;
-using ::executorch::runtime::internal::event_tracer_track_allocator;
-using ::executorch::runtime::internal::EventTracerProfileInstructionScope;
-using ::executorch::runtime::internal::EventTracerProfileMethodScope;
-using ::executorch::runtime::internal::EventTracerProfileOpScope;
-using ::executorch::runtime::internal::EventTracerProfileScope;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::
+    event_tracer_begin_profiling_event;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::
+    event_tracer_create_event_block;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::
+    event_tracer_end_profiling_event;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::event_tracer_log_evalue;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::
+    event_tracer_log_evalue_output;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::
+    event_tracer_set_bundled_input_index;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::
+    event_tracer_track_allocation;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::
+    event_tracer_track_allocator;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::
+    EventTracerProfileInstructionScope;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::
+    EventTracerProfileMethodScope;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::EventTracerProfileOpScope;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::EventTracerProfileScope;
 
 } // namespace internal
 } // namespace executor
diff --git a/runtime/core/exec_aten/exec_aten.h b/runtime/core/exec_aten/exec_aten.h
index 704bb868abd..10075ab5920 100644
--- a/runtime/core/exec_aten/exec_aten.h
+++ b/runtime/core/exec_aten/exec_aten.h
@@ -47,6 +47,21 @@
 
 #endif
 
+/**
+ * This hack is for separating out ATen mode vs non-ATen mode. In ATen mode,
+ * we use the ATen types directly. In non-ATen mode, we use the portable types.
+ * To avoid duplicate symbols and/or duplicate operator registration, when a
+ * user depends on both the ATen mode and non-ATen mode versions of the
+ * ExecuTorch library.
+ */
+#ifndef ET_RUNTIME_NAMESPACE
+#if defined(USE_ATEN_LIB)
+#define ET_RUNTIME_NAMESPACE runtime::aten
+#else
+#define ET_RUNTIME_NAMESPACE runtime
+#endif
+#endif
+
 namespace executorch {
 namespace aten {
 
diff --git a/runtime/core/exec_aten/testing_util/test/tensor_factory_test.cpp b/runtime/core/exec_aten/testing_util/test/tensor_factory_test.cpp
index ed8cc00f4ef..feb00f79b8f 100644
--- a/runtime/core/exec_aten/testing_util/test/tensor_factory_test.cpp
+++ b/runtime/core/exec_aten/testing_util/test/tensor_factory_test.cpp
@@ -26,8 +26,8 @@ using executorch::aten::SizesType;
 using executorch::aten::StridesType;
 using executorch::aten::Tensor;
 using executorch::aten::TensorList;
+using executorch::ET_RUNTIME_NAMESPACE::resize_tensor;
 using executorch::runtime::Error;
-using executorch::runtime::resize_tensor;
 using executorch::runtime::TensorShapeDynamism;
 using executorch::runtime::testing::TensorFactory;
 using executorch::runtime::testing::TensorListFactory;
diff --git a/runtime/core/exec_aten/util/tensor_util.h b/runtime/core/exec_aten/util/tensor_util.h
index 4e5a0cebb07..b0b79882361 100644
--- a/runtime/core/exec_aten/util/tensor_util.h
+++ b/runtime/core/exec_aten/util/tensor_util.h
@@ -396,8 +396,7 @@
       #scalar_tensor " could not be extracted: wrong type or out of range");
 
 namespace executorch {
-namespace runtime {
-
+namespace ET_RUNTIME_NAMESPACE {
 //
 // Utility functions for checking tensor attributes
 //
@@ -446,10 +445,10 @@ inline bool tensor_can_cast_to(
     executorch::aten::Tensor a,
     executorch::aten::ScalarType dtype) {
   ET_CHECK_OR_RETURN_FALSE(
-      torch::executor::canCast(a.scalar_type(), dtype),
+      ::torch::executor::canCast(a.scalar_type(), dtype),
       "Tensor of dtype %s cannot cast to dtype %s",
-      torch::executor::toString(a.scalar_type()),
-      torch::executor::toString(dtype));
+      ::torch::executor::toString(a.scalar_type()),
+      ::torch::executor::toString(dtype));
 
   return true;
 }
@@ -458,7 +457,7 @@ inline bool tensor_is_bool_type(executorch::aten::Tensor t) {
   ET_CHECK_OR_RETURN_FALSE(
       t.scalar_type() == executorch::aten::ScalarType::Bool,
       "Expected to find bool type, but tensor has type %s",
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
@@ -469,8 +468,8 @@ inline bool tensor_is_type(
   ET_CHECK_OR_RETURN_FALSE(
       t.scalar_type() == dtype,
       "Expected to find %s type, but tensor has type %s",
-      torch::executor::toString(dtype),
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(dtype),
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
@@ -482,9 +481,9 @@ inline bool tensor_is_type(
   ET_LOG_MSG_AND_RETURN_IF_FALSE(
       t.scalar_type() == dtype || t.scalar_type() == dtype2,
       "Expected to find %s or %s type, but tensor has type %s",
-      torch::executor::toString(dtype),
-      torch::executor::toString(dtype2),
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(dtype),
+      ::torch::executor::toString(dtype2),
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
@@ -498,10 +497,10 @@ inline bool tensor_is_type(
       t.scalar_type() == dtype || t.scalar_type() == dtype2 ||
           t.scalar_type() == dtype3,
       "Expected to find %s, %s, or %s type, but tensor has type %s",
-      torch::executor::toString(dtype),
-      torch::executor::toString(dtype2),
-      torch::executor::toString(dtype3),
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(dtype),
+      ::torch::executor::toString(dtype2),
+      ::torch::executor::toString(dtype3),
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
@@ -510,36 +509,36 @@ inline bool tensor_is_integral_type(
     executorch::aten::Tensor t,
     bool includeBool = false) {
   ET_CHECK_OR_RETURN_FALSE(
-      torch::executor::isIntegralType(t.scalar_type(), includeBool),
+      ::torch::executor::isIntegralType(t.scalar_type(), includeBool),
       "Expected to find a integral type, but tensor has type %s",
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
 
 inline bool tensor_is_floating_type(executorch::aten::Tensor t) {
   ET_CHECK_OR_RETURN_FALSE(
-      torch::executor::isFloatingType(t.scalar_type()),
+      ::torch::executor::isFloatingType(t.scalar_type()),
       "Expected to find a floating type, but tensor has type %s",
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
 
 inline bool tensor_is_real_type(executorch::aten::Tensor t) {
   ET_CHECK_OR_RETURN_FALSE(
-      torch::executor::isRealType(t.scalar_type()),
+      ::torch::executor::isRealType(t.scalar_type()),
       "Expected to find a real type, but tensor has type %s",
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
 
 inline bool tensor_is_realh_type(executorch::aten::Tensor t) {
   ET_CHECK_OR_RETURN_FALSE(
-      torch::executor::isRealHType(t.scalar_type()),
+      ::torch::executor::isRealHType(t.scalar_type()),
       "Expected to find a real type, but tensor has type %s",
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
@@ -548,16 +547,16 @@ inline bool tensor_is_realhbf16_type(executorch::aten::Tensor t) {
   ET_CHECK_OR_RETURN_FALSE(
       executorch::runtime::isRealHBF16Type(t.scalar_type()),
       "Expected to find a real type, but tensor has type %s",
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
 
 inline bool tensor_is_realhb_type(executorch::aten::Tensor t) {
   ET_CHECK_OR_RETURN_FALSE(
-      torch::executor::isRealHBType(t.scalar_type()),
+      ::torch::executor::isRealHBType(t.scalar_type()),
       "Expected to find a real type, but tensor has type %s",
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
@@ -566,25 +565,25 @@ inline bool tensor_is_realhbbf16_type(executorch::aten::Tensor t) {
   ET_CHECK_OR_RETURN_FALSE(
       executorch::runtime::isRealHBBF16Type(t.scalar_type()),
       "Expected to find a real type, but tensor has type %s",
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
 
 inline bool tensor_is_complex_type(executorch::aten::Tensor t) {
   ET_CHECK_OR_RETURN_FALSE(
-      torch::executor::isComplexType(t.scalar_type()),
+      ::torch::executor::isComplexType(t.scalar_type()),
       "Expected to find a complex type, but tensor has type %s",
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
 
 inline bool tensor_is_bits_type(executorch::aten::Tensor t) {
   ET_CHECK_OR_RETURN_FALSE(
-      torch::executor::isBitsType(t.scalar_type()),
+      ::torch::executor::isBitsType(t.scalar_type()),
       "Expected to find a bits type, but tensor has type %s",
-      torch::executor::toString(t.scalar_type()));
+      ::torch::executor::toString(t.scalar_type()));
 
   return true;
 }
@@ -595,8 +594,8 @@ inline bool tensors_have_same_dtype(
   ET_CHECK_OR_RETURN_FALSE(
       a.scalar_type() == b.scalar_type(),
       ET_TENSOR_CHECK_PREFIX__ ": dtype={%s, %s}",
-      torch::executor::toString(a.scalar_type()),
-      torch::executor::toString(b.scalar_type()));
+      ::torch::executor::toString(a.scalar_type()),
+      ::torch::executor::toString(b.scalar_type()));
   return true;
 }
 
@@ -607,9 +606,9 @@ inline bool tensors_have_same_dtype(
   ET_CHECK_OR_RETURN_FALSE(
       a.scalar_type() == b.scalar_type() && b.scalar_type() == c.scalar_type(),
       ET_TENSOR_CHECK_PREFIX__ ": dtype={%s, %s, %s}",
-      torch::executor::toString(a.scalar_type()),
-      torch::executor::toString(b.scalar_type()),
-      torch::executor::toString(c.scalar_type()));
+      ::torch::executor::toString(a.scalar_type()),
+      ::torch::executor::toString(b.scalar_type()),
+      ::torch::executor::toString(c.scalar_type()));
   return true;
 }
 
@@ -1349,60 +1348,61 @@ inline size_t calculate_linear_index(
   return index;
 }
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 namespace torch {
 namespace executor {
 // TODO(T197294990): Remove these deprecated aliases once all users have moved
 // to the new `::executorch` namespaces.
-using ::executorch::runtime::calculate_linear_index;
-using ::executorch::runtime::coordinateToIndex;
-using ::executorch::runtime::dim_is_valid;
-using ::executorch::runtime::extract_scalar_tensor;
-using ::executorch::runtime::get_dim_order;
-using ::executorch::runtime::getLeadingDims;
-using ::executorch::runtime::getTrailingDims;
-using ::executorch::runtime::indexToCoordinate;
+using ::executorch::ET_RUNTIME_NAMESPACE::calculate_linear_index;
+using ::executorch::ET_RUNTIME_NAMESPACE::coordinateToIndex;
+using ::executorch::ET_RUNTIME_NAMESPACE::dim_is_valid;
+using ::executorch::ET_RUNTIME_NAMESPACE::extract_scalar_tensor;
+using ::executorch::ET_RUNTIME_NAMESPACE::get_dim_order;
+using ::executorch::ET_RUNTIME_NAMESPACE::getLeadingDims;
+using ::executorch::ET_RUNTIME_NAMESPACE::getTrailingDims;
+using ::executorch::ET_RUNTIME_NAMESPACE::indexToCoordinate;
+using ::executorch::ET_RUNTIME_NAMESPACE::nonempty_size;
+using ::executorch::ET_RUNTIME_NAMESPACE::nonzero_dim;
+using ::executorch::ET_RUNTIME_NAMESPACE::resize;
+using ::executorch::ET_RUNTIME_NAMESPACE::resize_tensor;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_can_cast_to;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_dim_has_index;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_has_dim;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_has_expected_size;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_has_non_empty_dim;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_has_rank_greater_or_equal_to;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_has_rank_smaller_or_equal_to;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_has_valid_dim_order;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_bits_type;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_bool_type;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_complex_type;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_contiguous;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_default_dim_order;
+using ::executorch::ET_RUNTIME_NAMESPACE::
+    tensor_is_default_or_channels_last_dim_order;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_floating_type;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_integral_type;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_rank;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_real_type;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_realh_type;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_realhb_type;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensor_is_scalar;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_dim_order;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_dtype;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_rank;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_shape;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_shape_and_dtype;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_size_at_dims;
+using ::executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_strides;
 using ::executorch::runtime::kTensorDimensionLimit;
-using ::executorch::runtime::nonempty_size;
-using ::executorch::runtime::nonzero_dim;
-using ::executorch::runtime::resize;
-using ::executorch::runtime::resize_tensor;
-using ::executorch::runtime::tensor_can_cast_to;
-using ::executorch::runtime::tensor_dim_has_index;
-using ::executorch::runtime::tensor_has_dim;
-using ::executorch::runtime::tensor_has_expected_size;
-using ::executorch::runtime::tensor_has_non_empty_dim;
-using ::executorch::runtime::tensor_has_rank_greater_or_equal_to;
-using ::executorch::runtime::tensor_has_rank_smaller_or_equal_to;
-using ::executorch::runtime::tensor_has_valid_dim_order;
-using ::executorch::runtime::tensor_is_bits_type;
-using ::executorch::runtime::tensor_is_bool_type;
-using ::executorch::runtime::tensor_is_complex_type;
-using ::executorch::runtime::tensor_is_contiguous;
-using ::executorch::runtime::tensor_is_default_dim_order;
-using ::executorch::runtime::tensor_is_default_or_channels_last_dim_order;
-using ::executorch::runtime::tensor_is_floating_type;
-using ::executorch::runtime::tensor_is_integral_type;
-using ::executorch::runtime::tensor_is_rank;
-using ::executorch::runtime::tensor_is_real_type;
-using ::executorch::runtime::tensor_is_realh_type;
-using ::executorch::runtime::tensor_is_realhb_type;
-using ::executorch::runtime::tensor_is_scalar;
-using ::executorch::runtime::tensors_have_same_dim_order;
-using ::executorch::runtime::tensors_have_same_dtype;
-using ::executorch::runtime::tensors_have_same_rank;
-using ::executorch::runtime::tensors_have_same_shape;
-using ::executorch::runtime::tensors_have_same_shape_and_dtype;
-using ::executorch::runtime::tensors_have_same_size_at_dims;
-using ::executorch::runtime::tensors_have_same_strides;
 namespace internal {
-using ::executorch::runtime::internal::copy_tensor_data;
-using ::executorch::runtime::internal::reset_data_ptr;
-using ::executorch::runtime::internal::resize_tensor_impl;
-using ::executorch::runtime::internal::set_tensor_data;
-using ::executorch::runtime::internal::share_tensor_data;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::copy_tensor_data;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::reset_data_ptr;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::resize_tensor_impl;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::set_tensor_data;
+using ::executorch::ET_RUNTIME_NAMESPACE::internal::share_tensor_data;
 } // namespace internal
 } // namespace executor
 } // namespace torch
diff --git a/runtime/core/exec_aten/util/tensor_util_aten.cpp b/runtime/core/exec_aten/util/tensor_util_aten.cpp
index 4df273d4dbb..ddfd0560a69 100644
--- a/runtime/core/exec_aten/util/tensor_util_aten.cpp
+++ b/runtime/core/exec_aten/util/tensor_util_aten.cpp
@@ -12,7 +12,7 @@
 #include <executorch/runtime/platform/assert.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 /**
  * Implementation for ATen tensor util, should only be included in
  * `<target>_aten` target and only be used in ATen mode. Explicitly taking
@@ -214,6 +214,5 @@ Error resize_tensor_impl(
 }
 
 } // namespace internal
-
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/core/exec_aten/util/test/tensor_util_test.cpp b/runtime/core/exec_aten/util/test/tensor_util_test.cpp
index 7d30b0bbdbe..cdc391adf20 100644
--- a/runtime/core/exec_aten/util/test/tensor_util_test.cpp
+++ b/runtime/core/exec_aten/util/test/tensor_util_test.cpp
@@ -17,7 +17,7 @@
 using namespace ::testing;
 using executorch::aten::ScalarType;
 using executorch::aten::Tensor;
-using executorch::runtime::extract_scalar_tensor;
+using executorch::ET_RUNTIME_NAMESPACE::extract_scalar_tensor;
 using executorch::runtime::testing::TensorFactory;
 
 class TensorUtilTest : public ::testing::Test {
@@ -148,13 +148,13 @@ TEST_F(TensorUtilTest, GetLeadingDimsSmokeTest) {
   Tensor t = tf_int_.ones({2, 3, 4});
 
   // getLeadingDims(t, 1) => t.size(0)
-  EXPECT_EQ(executorch::runtime::getLeadingDims(t, 1), 2);
+  EXPECT_EQ(executorch::ET_RUNTIME_NAMESPACE::getLeadingDims(t, 1), 2);
 
   // getLeadingDims(t, 2) => t.size(0) * t.size(1)
-  EXPECT_EQ(executorch::runtime::getLeadingDims(t, 2), 6);
+  EXPECT_EQ(executorch::ET_RUNTIME_NAMESPACE::getLeadingDims(t, 2), 6);
 
   // getLeadingDims(t, 3) => t.size(0) * t.size(1) * t.size(2)
-  EXPECT_EQ(executorch::runtime::getLeadingDims(t, 3), 24);
+  EXPECT_EQ(executorch::ET_RUNTIME_NAMESPACE::getLeadingDims(t, 3), 24);
 }
 
 TEST_F(TensorUtilTest, GetLeadingDimsInputOutOfBoundDies) {
@@ -162,9 +162,9 @@ TEST_F(TensorUtilTest, GetLeadingDimsInputOutOfBoundDies) {
   Tensor t = tf_int_.ones({2, 3, 4});
 
   // dim needs to be in the range [0, t.dim()]
-  ET_EXPECT_DEATH(executorch::runtime::getLeadingDims(t, -2), "");
-  ET_EXPECT_DEATH(executorch::runtime::getLeadingDims(t, -1), "");
-  ET_EXPECT_DEATH(executorch::runtime::getLeadingDims(t, 4), "");
+  ET_EXPECT_DEATH(executorch::ET_RUNTIME_NAMESPACE::getLeadingDims(t, -2), "");
+  ET_EXPECT_DEATH(executorch::ET_RUNTIME_NAMESPACE::getLeadingDims(t, -1), "");
+  ET_EXPECT_DEATH(executorch::ET_RUNTIME_NAMESPACE::getLeadingDims(t, 4), "");
 }
 
 TEST_F(TensorUtilTest, GetTrailingDimsSmokeTest) {
@@ -172,13 +172,13 @@ TEST_F(TensorUtilTest, GetTrailingDimsSmokeTest) {
   Tensor t = tf_int_.ones({2, 3, 4});
 
   // getTrailingDims(t, 1) => t.size(2)
-  EXPECT_EQ(executorch::runtime::getTrailingDims(t, 1), 4);
+  EXPECT_EQ(executorch::ET_RUNTIME_NAMESPACE::getTrailingDims(t, 1), 4);
 
   // getTrailingDims(t, 0) => t.size(1) * t.size(2)
-  EXPECT_EQ(executorch::runtime::getTrailingDims(t, 0), 12);
+  EXPECT_EQ(executorch::ET_RUNTIME_NAMESPACE::getTrailingDims(t, 0), 12);
 
   // getTrailingDims(t, -1) => t.size(0) * t.size(1) * t.size(2)
-  EXPECT_EQ(executorch::runtime::getTrailingDims(t, -1), 24);
+  EXPECT_EQ(executorch::ET_RUNTIME_NAMESPACE::getTrailingDims(t, -1), 24);
 }
 
 TEST_F(TensorUtilTest, GetTrailingDimsInputOutOfBoundDies) {
@@ -186,9 +186,9 @@ TEST_F(TensorUtilTest, GetTrailingDimsInputOutOfBoundDies) {
   Tensor t = tf_int_.ones({2, 3, 4});
 
   // dim needs to be in the range [-1, t.dim() - 1)
-  ET_EXPECT_DEATH(executorch::runtime::getTrailingDims(t, -2), "");
-  ET_EXPECT_DEATH(executorch::runtime::getTrailingDims(t, 3), "");
-  ET_EXPECT_DEATH(executorch::runtime::getTrailingDims(t, 4), "");
+  ET_EXPECT_DEATH(executorch::ET_RUNTIME_NAMESPACE::getTrailingDims(t, -2), "");
+  ET_EXPECT_DEATH(executorch::ET_RUNTIME_NAMESPACE::getTrailingDims(t, 3), "");
+  ET_EXPECT_DEATH(executorch::ET_RUNTIME_NAMESPACE::getTrailingDims(t, 4), "");
 }
 
 TEST_F(TensorUtilTest, ContiguousCheckSupported) {
@@ -421,7 +421,7 @@ TEST_F(TensorUtilTest, BoolTensorNotScalarFails) {
 //
 
 TEST_F(TensorUtilTest, TensorIsRankTest) {
-  using executorch::runtime::tensor_is_rank;
+  using executorch::ET_RUNTIME_NAMESPACE::tensor_is_rank;
   Tensor a = tf_float_.ones({2, 3, 5});
 
   EXPECT_TRUE(tensor_is_rank(a, 3));
@@ -430,7 +430,7 @@ TEST_F(TensorUtilTest, TensorIsRankTest) {
 }
 
 TEST_F(TensorUtilTest, TensorHasDimTest) {
-  using executorch::runtime::tensor_has_dim;
+  using executorch::ET_RUNTIME_NAMESPACE::tensor_has_dim;
   Tensor a = tf_float_.ones({2, 3, 5});
 
   EXPECT_TRUE(tensor_has_dim(a, 2));
@@ -445,7 +445,7 @@ TEST_F(TensorUtilTest, TensorHasDimTest) {
 }
 
 TEST_F(TensorUtilTest, TensorsHaveSameDtypeTest) {
-  using executorch::runtime::tensors_have_same_dtype;
+  using executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_dtype;
   Tensor a = tf_float_.ones({2, 3});
   Tensor b = tf_float_.ones({2, 3});
   Tensor c = tf_float_.ones({3, 3});
@@ -458,7 +458,7 @@ TEST_F(TensorUtilTest, TensorsHaveSameDtypeTest) {
 }
 
 TEST_F(TensorUtilTest, TensorsHaveSameSizeAtDimTest) {
-  using executorch::runtime::tensors_have_same_size_at_dims;
+  using executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_size_at_dims;
   Tensor a = tf_float_.ones({2, 3, 4, 5});
   Tensor b = tf_float_.ones({5, 4, 3, 2});
 
@@ -470,7 +470,7 @@ TEST_F(TensorUtilTest, TensorsHaveSameSizeAtDimTest) {
 }
 
 TEST_F(TensorUtilTest, TensorsHaveSameShapeTest) {
-  using executorch::runtime::tensors_have_same_shape;
+  using executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_shape;
   Tensor a = tf_float_.ones({2, 3});
   Tensor b = tf_int_.ones({2, 3});
   Tensor c = tf_byte_.ones({2, 3});
@@ -493,7 +493,7 @@ TEST_F(TensorUtilTest, TensorsHaveSameShapeTest) {
 }
 
 TEST_F(TensorUtilTest, TensorsHaveSameShapeAndDtypeTest) {
-  using executorch::runtime::tensors_have_same_shape_and_dtype;
+  using executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_shape_and_dtype;
   Tensor a = tf_float_.ones({2, 3});
   Tensor b = tf_float_.ones({2, 3});
   Tensor c = tf_float_.ones({2, 3});
@@ -515,7 +515,7 @@ TEST_F(TensorUtilTest, TensorsHaveSameShapeAndDtypeTest) {
 }
 
 TEST_F(TensorUtilTest, TensorsHaveSameStridesTest) {
-  using executorch::runtime::tensors_have_same_strides;
+  using executorch::ET_RUNTIME_NAMESPACE::tensors_have_same_strides;
   Tensor a = tf_float_.full_channels_last({4, 5, 2, 3}, 1);
   Tensor b = tf_float_.full_channels_last({4, 5, 2, 3}, 2);
   Tensor c = tf_float_.full_channels_last({4, 5, 2, 3}, 3);
@@ -530,7 +530,7 @@ TEST_F(TensorUtilTest, TensorsHaveSameStridesTest) {
 }
 
 TEST_F(TensorUtilTest, TensorIsContiguous) {
-  using executorch::runtime::tensor_is_contiguous;
+  using executorch::ET_RUNTIME_NAMESPACE::tensor_is_contiguous;
   // Note that the strides.size() == 0 case is not tested, since
   Tensor a = tf_float_.full_channels_last({4, 5, 2, 3}, 1);
   Tensor b = tf_float_.ones({4, 5, 2, 3});
@@ -547,7 +547,7 @@ TEST_F(TensorUtilTest, ResizeZeroDimTensor) {
   Tensor a = tf_float_.ones({});
 
   EXPECT_EQ(
-      executorch::runtime::resize_tensor(a, {}),
+      executorch::ET_RUNTIME_NAMESPACE::resize_tensor(a, {}),
       executorch::runtime::Error::Ok);
   EXPECT_EQ(a.dim(), 0);
 }
diff --git a/runtime/core/named_data_map.h b/runtime/core/named_data_map.h
index e79c7035989..14179d22795 100644
--- a/runtime/core/named_data_map.h
+++ b/runtime/core/named_data_map.h
@@ -22,8 +22,7 @@
 #include <executorch/runtime/platform/compiler.h>
 
 namespace executorch {
-namespace runtime {
-
+namespace ET_RUNTIME_NAMESPACE {
 /**
  * Interface to access and retrieve data via name.
  * See executorch/extension/flat_tensor/ for an example.
@@ -37,8 +36,8 @@ class ET_EXPERIMENTAL NamedDataMap {
    * @param key The name of the tensor.
    * @return Result containing TensorLayout with tensor metadata.
    */
-  ET_NODISCARD virtual Result<const executorch::runtime::TensorLayout>
-  get_metadata(const char* key) const = 0;
+  ET_NODISCARD virtual Result<const TensorLayout> get_metadata(
+      const char* key) const = 0;
   /**
    * Get data by key.
    *
@@ -78,7 +77,7 @@ class ET_EXPERIMENTAL NamedDataMap {
   ET_NODISCARD virtual Result<const char*> get_key(size_t index) const = 0;
 };
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 #ifdef __GNUC__
diff --git a/runtime/core/targets.bzl b/runtime/core/targets.bzl
index 3195e727d96..d3e02b1afb5 100644
--- a/runtime/core/targets.bzl
+++ b/runtime/core/targets.bzl
@@ -95,9 +95,9 @@ def define_common_targets():
                 "@EXECUTORCH_CLIENTS",
             ],
             exported_deps = [
-                "//executorch/runtime/core:core",
-                "//executorch/runtime/core/exec_aten:lib" + aten_suffix,
+                ":core",
                 ":tag",
+                "//executorch/runtime/core/exec_aten:lib" + aten_suffix,
             ],
         )
 
@@ -119,6 +119,37 @@ def define_common_targets():
             ],
         )
 
+        runtime.cxx_library(
+            name = "named_data_map" + aten_suffix,
+            exported_headers = [
+                "named_data_map.h",
+            ],
+            visibility = [
+                "//executorch/...",
+                "@EXECUTORCH_CLIENTS",
+            ],
+            exported_deps = [
+                ":tensor_layout" + aten_suffix,
+                "//executorch/runtime/core/exec_aten:lib" + aten_suffix,
+            ],
+        )
+
+
+        runtime.cxx_library(
+            name = "tensor_layout" + aten_suffix,
+            srcs = ["tensor_layout.cpp"],
+            exported_headers = ["tensor_layout.h"],
+            deps = [
+                "//executorch/runtime/core/portable_type/c10/c10:c10",
+            ],
+            exported_deps = [
+                ":core",
+                "//executorch/runtime/core/exec_aten:lib" + aten_suffix,
+                "//executorch/runtime/core/exec_aten/util:scalar_type_util" + aten_suffix,
+            ],
+            visibility = ["//executorch/..."],
+        )
+
     runtime.cxx_library(
         name = "tag",
         srcs = ["tag.cpp"],
@@ -133,31 +164,3 @@ def define_common_targets():
             "//executorch/...",
         ],
     )
-
-    runtime.cxx_library(
-        name = "named_data_map",
-        exported_headers = [
-            "named_data_map.h",
-        ],
-        visibility = [
-            "//executorch/...",
-            "@EXECUTORCH_CLIENTS",
-        ],
-        exported_deps = [
-            ":tensor_layout",
-        ],
-    )
-
-    runtime.cxx_library(
-        name = "tensor_layout",
-        srcs = ["tensor_layout.cpp"],
-        exported_headers = ["tensor_layout.h"],
-        deps = [
-            "//executorch/runtime/core/portable_type/c10/c10:c10",
-        ],
-        exported_deps = [
-            ":core",
-            "//executorch/runtime/core/exec_aten:lib",
-        ],
-        visibility = ["//executorch/..."],
-    )
diff --git a/runtime/core/tensor_layout.cpp b/runtime/core/tensor_layout.cpp
index 2b862e6dc14..d33f79f27c4 100644
--- a/runtime/core/tensor_layout.cpp
+++ b/runtime/core/tensor_layout.cpp
@@ -13,7 +13,7 @@
 #include <executorch/runtime/core/tensor_layout.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
 namespace {
 Result<size_t> calculate_nbytes(
@@ -51,5 +51,5 @@ Result<const TensorLayout> TensorLayout::create(
   }
   return TensorLayout(sizes, dim_order, scalar_type, nbytes.get());
 }
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/core/tensor_layout.h b/runtime/core/tensor_layout.h
index c2c3833f528..42131e6506e 100644
--- a/runtime/core/tensor_layout.h
+++ b/runtime/core/tensor_layout.h
@@ -14,7 +14,7 @@
 #include <executorch/runtime/core/span.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
 /**
  * Describes the layout of a tensor.
@@ -89,5 +89,5 @@ class ET_EXPERIMENTAL TensorLayout final {
   const size_t nbytes_;
 };
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/executor/method.cpp b/runtime/executor/method.cpp
index 41d44522a22..63a3166bfdd 100644
--- a/runtime/executor/method.cpp
+++ b/runtime/executor/method.cpp
@@ -32,9 +32,8 @@
 #include <executorch/schema/program_generated.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
-using deserialization::NamedData;
 using internal::PlatformMemoryAllocator;
 
 /**
@@ -1640,5 +1639,5 @@ Method::~Method() {
   }
   // All other fields are trivially destructible.
 }
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/executor/method.h b/runtime/executor/method.h
index 0ca2df440ad..c730af5778b 100644
--- a/runtime/executor/method.h
+++ b/runtime/executor/method.h
@@ -32,7 +32,7 @@ struct EValue;
 } // namespace executorch_flatbuffer
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
 // Forward declare NamedData. This is a public header and must not include
 // internal data types.
@@ -394,14 +394,14 @@ class Method final {
   void log_outputs();
 };
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 namespace torch {
 namespace executor {
 // TODO(T197294990): Remove these deprecated aliases once all users have moved
 // to the new `::executorch` namespaces.
-using ::executorch::runtime::Method;
+using ::executorch::ET_RUNTIME_NAMESPACE::Method;
 } // namespace executor
 } // namespace torch
 
diff --git a/runtime/executor/method_meta.cpp b/runtime/executor/method_meta.cpp
index 8f84fea940f..687b9c30471 100644
--- a/runtime/executor/method_meta.cpp
+++ b/runtime/executor/method_meta.cpp
@@ -16,7 +16,7 @@
 #include <executorch/schema/program_generated.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
 namespace {
 Result<Tag> get_tag(
@@ -279,6 +279,5 @@ size_t MethodMeta::num_instructions() const {
   }
   return num_instructions;
 }
-
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/executor/method_meta.h b/runtime/executor/method_meta.h
index d9bb64d68a7..a74cb090523 100644
--- a/runtime/executor/method_meta.h
+++ b/runtime/executor/method_meta.h
@@ -20,7 +20,7 @@ struct ExecutionPlan;
 } // namespace executorch_flatbuffer
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
 /**
  * Metadata about a specific tensor of an ExecuTorch Program.
@@ -240,14 +240,14 @@ class MethodMeta final {
   const executorch_flatbuffer::ExecutionPlan* s_plan_;
 };
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 namespace torch {
 namespace executor {
 // TODO(T197294990): Remove these deprecated aliases once all users have moved
 // to the new `::executorch` namespaces.
-using ::executorch::runtime::MethodMeta;
-using ::executorch::runtime::TensorInfo;
+using ::executorch::ET_RUNTIME_NAMESPACE::MethodMeta;
+using ::executorch::ET_RUNTIME_NAMESPACE::TensorInfo;
 } // namespace executor
 } // namespace torch
diff --git a/runtime/executor/platform_memory_allocator.h b/runtime/executor/platform_memory_allocator.h
index 09195a460ac..7ab58bf0f3d 100644
--- a/runtime/executor/platform_memory_allocator.h
+++ b/runtime/executor/platform_memory_allocator.h
@@ -17,7 +17,7 @@
 #include <executorch/runtime/platform/platform.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 namespace internal {
 
 /**
@@ -107,5 +107,5 @@ class PlatformMemoryAllocator final : public MemoryAllocator {
 };
 
 } // namespace internal
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/executor/program.cpp b/runtime/executor/program.cpp
index 14e0b83d8aa..238c806b1d6 100644
--- a/runtime/executor/program.cpp
+++ b/runtime/executor/program.cpp
@@ -28,8 +28,7 @@
 #endif
 
 namespace executorch {
-namespace runtime {
-
+namespace ET_RUNTIME_NAMESPACE {
 namespace {
 
 /**
@@ -535,5 +534,5 @@ Error Program::load_mutable_subsegment_into(
       segment_base_offset_ + segment->offset() + offset, size, info, buffer);
 }
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/executor/program.h b/runtime/executor/program.h
index 0932e51619f..9670fd7c79f 100644
--- a/runtime/executor/program.h
+++ b/runtime/executor/program.h
@@ -36,8 +36,7 @@ struct Program;
 } // namespace executorch_flatbuffer
 
 namespace executorch {
-namespace runtime {
-
+namespace ET_RUNTIME_NAMESPACE {
 namespace testing {
 // Provides test access to private Program methods.
 class ProgramTestFriend;
@@ -313,14 +312,14 @@ class Program final {
   std::optional<internal::PteDataMap> pte_data_map_;
 };
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 namespace torch {
 namespace executor {
 // TODO(T197294990): Remove these deprecated aliases once all users have moved
 // to the new `::executorch` namespaces.
-using ::executorch::runtime::Program;
+using ::executorch::ET_RUNTIME_NAMESPACE::Program;
 } // namespace executor
 } // namespace torch
 
diff --git a/runtime/executor/pte_data_map.cpp b/runtime/executor/pte_data_map.cpp
index 5829395028a..fd064cb8256 100644
--- a/runtime/executor/pte_data_map.cpp
+++ b/runtime/executor/pte_data_map.cpp
@@ -10,7 +10,7 @@
 #include <executorch/schema/program_generated.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 namespace internal {
 
 /* static */ executorch::runtime::Result<PteDataMap> PteDataMap::create(
@@ -83,5 +83,5 @@ ET_NODISCARD executorch::runtime::Result<const char*> PteDataMap::get_key(
 }
 
 } // namespace internal
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/executor/pte_data_map.h b/runtime/executor/pte_data_map.h
index 01c15555786..b26c0ac42f9 100644
--- a/runtime/executor/pte_data_map.h
+++ b/runtime/executor/pte_data_map.h
@@ -46,7 +46,7 @@ using FlatbufferDataSegment = flatbuffers::
 #endif
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 namespace internal {
 
 /**
@@ -147,5 +147,5 @@ class PteDataMap final : public NamedDataMap {
 };
 
 } // namespace internal
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/executor/targets.bzl b/runtime/executor/targets.bzl
index cfb6c607359..649b2c13cc1 100644
--- a/runtime/executor/targets.bzl
+++ b/runtime/executor/targets.bzl
@@ -42,30 +42,33 @@ def define_common_targets():
         ],
     )
 
-    runtime.cxx_library(
-        name = "pte_data_map",
-        srcs = [
-            "pte_data_map.cpp",
-        ],
-        exported_headers = [
-            "pte_data_map.h",
-        ],
-        visibility = [
-            "//executorch/runtime/executor/...",
-            "@EXECUTORCH_CLIENTS",
-        ],
-        exported_deps = [
-            "//executorch/runtime/core:core",
-            "//executorch/runtime/core:named_data_map",
-        ],
-        deps = [
-            "//executorch/schema:program",
-        ],
-        exported_preprocessor_flags = [] if runtime.is_oss else ["-DEXECUTORCH_INTERNAL_FLATBUFFERS=1"],
-    )
 
     for aten_mode in get_aten_mode_options():
         aten_suffix = "_aten" if aten_mode else ""
+
+        runtime.cxx_library(
+            name = "pte_data_map" + aten_suffix,
+            srcs = [
+                "pte_data_map.cpp",
+            ],
+            exported_headers = [
+                "pte_data_map.h",
+            ],
+            visibility = [
+                "//executorch/runtime/executor/...",
+                "@EXECUTORCH_CLIENTS",
+            ],
+            exported_deps = [
+                "//executorch/runtime/core:core",
+                "//executorch/runtime/core:named_data_map" + aten_suffix,
+                "//executorch/runtime/core/exec_aten/util:scalar_type_util" + aten_suffix,
+            ],
+            deps = [
+                "//executorch/schema:program",
+            ],
+            exported_preprocessor_flags = [] if runtime.is_oss else ["-DEXECUTORCH_INTERNAL_FLATBUFFERS=1"],
+        )
+
         runtime.cxx_library(
             name = "program" + aten_suffix,
             exported_deps = [
@@ -103,17 +106,17 @@ def define_common_targets():
             preprocessor_flags = _program_preprocessor_flags(),
             exported_deps = [
                 ":memory_manager",
-                ":pte_data_map",
-                "//executorch/runtime/backend:interface",
+                ":pte_data_map" + aten_suffix,
+                "//executorch/runtime/backend:interface" + aten_suffix,
                 "//executorch/runtime/core:core",
-                "//executorch/runtime/core:named_data_map",
+                "//executorch/runtime/core:named_data_map" + aten_suffix,
                 "//executorch/runtime/core:evalue" + aten_suffix,
                 "//executorch/runtime/core:event_tracer" + aten_suffix,
                 "//executorch/runtime/core/exec_aten:lib" + aten_suffix,
                 "//executorch/runtime/core/exec_aten/util:scalar_type_util" + aten_suffix,
                 "//executorch/runtime/core/exec_aten/util:tensor_util" + aten_suffix,
                 "//executorch/runtime/kernel:kernel_runtime_context" + aten_suffix,
-                "//executorch/runtime/kernel:operator_registry",
+                "//executorch/runtime/kernel:operator_registry" + aten_suffix,
                 "//executorch/runtime/platform:platform",
                 "//executorch/schema:extended_header",
             ],
diff --git a/runtime/executor/tensor_parser.h b/runtime/executor/tensor_parser.h
index 1fae84cfb05..e2b5ff8d6ea 100644
--- a/runtime/executor/tensor_parser.h
+++ b/runtime/executor/tensor_parser.h
@@ -21,7 +21,7 @@
 #include <executorch/schema/program_generated.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 namespace deserialization {
 
 /// Data structure to hold key and data buffer for external data used
@@ -142,7 +142,7 @@ ET_NODISCARD Result<void*> getTensorDataPtr(
     Span<NamedData> external_constants = {});
 
 } // namespace deserialization
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 namespace torch {
@@ -150,10 +150,11 @@ namespace executor {
 namespace deserialization {
 // TODO(T197294990): Remove these deprecated aliases once all users have moved
 // to the new `::executorch` namespaces.
-using ::executorch::runtime::deserialization::getTensorDataPtr;
-using ::executorch::runtime::deserialization::parseListOptionalType;
-using ::executorch::runtime::deserialization::parseTensor;
-using ::executorch::runtime::deserialization::parseTensorList;
+using ::executorch::ET_RUNTIME_NAMESPACE::deserialization::getTensorDataPtr;
+using ::executorch::ET_RUNTIME_NAMESPACE::deserialization::
+    parseListOptionalType;
+using ::executorch::ET_RUNTIME_NAMESPACE::deserialization::parseTensor;
+using ::executorch::ET_RUNTIME_NAMESPACE::deserialization::parseTensorList;
 } // namespace deserialization
 } // namespace executor
 } // namespace torch
diff --git a/runtime/executor/tensor_parser_aten.cpp b/runtime/executor/tensor_parser_aten.cpp
index d1a2f712853..2d454d15be5 100644
--- a/runtime/executor/tensor_parser_aten.cpp
+++ b/runtime/executor/tensor_parser_aten.cpp
@@ -19,7 +19,9 @@
 #include <ATen/ATen.h> // @donotremove @manual=//caffe2/aten:ATen-core
 
 namespace executorch {
+// This file is only used in ATen mode, so we use the runtime_aten namespace.
 namespace runtime {
+namespace aten {
 namespace deserialization {
 
 namespace {
@@ -126,5 +128,6 @@ Result<at::Tensor> parseTensor(
 }
 
 } // namespace deserialization
+} // namespace aten
 } // namespace runtime
 } // namespace executorch
diff --git a/runtime/executor/tensor_parser_exec_aten.cpp b/runtime/executor/tensor_parser_exec_aten.cpp
index 14ba5e0d42c..aa27bbf929d 100644
--- a/runtime/executor/tensor_parser_exec_aten.cpp
+++ b/runtime/executor/tensor_parser_exec_aten.cpp
@@ -16,11 +16,10 @@
 #include <executorch/schema/program_generated.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 namespace deserialization {
 
 using executorch::aten::ScalarType;
-using executorch::runtime::TensorLayout;
 // Provides access to private Program methods.
 class TensorParser final {
  public:
@@ -256,5 +255,5 @@ ET_NODISCARD Result<void*> getTensorDataPtr(
 }
 
 } // namespace deserialization
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/executor/tensor_parser_portable.cpp b/runtime/executor/tensor_parser_portable.cpp
index 787af8b506b..e1f09d557ac 100644
--- a/runtime/executor/tensor_parser_portable.cpp
+++ b/runtime/executor/tensor_parser_portable.cpp
@@ -18,13 +18,13 @@
 #include <executorch/schema/program_generated.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 namespace deserialization {
 
 using executorch::runtime::Span;
-using torch::executor::ScalarType;
-using torch::executor::Tensor;
-using torch::executor::TensorImpl;
+using ::torch::executor::ScalarType;
+using ::torch::executor::Tensor;
+using ::torch::executor::TensorImpl;
 
 Result<Tensor> parseTensor(
     const Program* program,
@@ -176,5 +176,5 @@ Result<Tensor> parseTensor(
 }
 
 } // namespace deserialization
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/executor/test/executor_test.cpp b/runtime/executor/test/executor_test.cpp
index 328b23a8df3..e2a44429941 100644
--- a/runtime/executor/test/executor_test.cpp
+++ b/runtime/executor/test/executor_test.cpp
@@ -22,14 +22,14 @@ using executorch::aten::Scalar;
 using executorch::aten::ScalarType;
 using executorch::aten::SizesType;
 using executorch::aten::Tensor;
+using executorch::ET_RUNTIME_NAMESPACE::get_op_function_from_registry;
+using executorch::ET_RUNTIME_NAMESPACE::Kernel;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
+using executorch::ET_RUNTIME_NAMESPACE::OpFunction;
+using executorch::ET_RUNTIME_NAMESPACE::register_kernel;
+using executorch::ET_RUNTIME_NAMESPACE::registry_has_op_function;
 using executorch::runtime::Error;
 using executorch::runtime::EValue;
-using executorch::runtime::get_op_function_from_registry;
-using executorch::runtime::Kernel;
-using executorch::runtime::KernelRuntimeContext;
-using executorch::runtime::OpFunction;
-using executorch::runtime::register_kernel;
-using executorch::runtime::registry_has_op_function;
 using executorch::runtime::Result;
 using executorch::runtime::testing::TensorFactory;
 
diff --git a/runtime/executor/test/targets.bzl b/runtime/executor/test/targets.bzl
index dd5262b5ac6..940601779cb 100644
--- a/runtime/executor/test/targets.bzl
+++ b/runtime/executor/test/targets.bzl
@@ -70,7 +70,7 @@ def define_common_targets(is_fbcode = False):
                 "//executorch/runtime/core/exec_aten:lib" + aten_suffix,
                 "//executorch/runtime/core:evalue" + aten_suffix,
                 "//executorch/runtime/kernel:kernel_runtime_context" + aten_suffix,
-                "//executorch/runtime/kernel:operator_registry",
+                "//executorch/runtime/kernel:operator_registry" + aten_suffix,
                 "//executorch/runtime/platform:platform",
             ],
         )
diff --git a/runtime/executor/test/test_backend_compiler_lib.cpp b/runtime/executor/test/test_backend_compiler_lib.cpp
index 9eea6384d6f..ce631eb4f57 100644
--- a/runtime/executor/test/test_backend_compiler_lib.cpp
+++ b/runtime/executor/test/test_backend_compiler_lib.cpp
@@ -13,13 +13,13 @@
 #include <cstdio>
 #include <cstdlib> /* strtol */
 
+using executorch::ET_RUNTIME_NAMESPACE::Backend;
+using executorch::ET_RUNTIME_NAMESPACE::BackendExecutionContext;
+using executorch::ET_RUNTIME_NAMESPACE::BackendInitContext;
+using executorch::ET_RUNTIME_NAMESPACE::BackendInterface;
+using executorch::ET_RUNTIME_NAMESPACE::CompileSpec;
+using executorch::ET_RUNTIME_NAMESPACE::DelegateHandle;
 using executorch::runtime::ArrayRef;
-using executorch::runtime::Backend;
-using executorch::runtime::BackendExecutionContext;
-using executorch::runtime::BackendInitContext;
-using executorch::runtime::BackendInterface;
-using executorch::runtime::CompileSpec;
-using executorch::runtime::DelegateHandle;
 using executorch::runtime::Error;
 using executorch::runtime::EValue;
 using executorch::runtime::FreeableBuffer;
diff --git a/runtime/executor/test/test_backend_with_delegate_mapping.cpp b/runtime/executor/test/test_backend_with_delegate_mapping.cpp
index e6d84aca189..a0b79b09c6d 100644
--- a/runtime/executor/test/test_backend_with_delegate_mapping.cpp
+++ b/runtime/executor/test/test_backend_with_delegate_mapping.cpp
@@ -14,13 +14,13 @@
 #include <cstdlib> /* strtol */
 #include <cstring>
 
+using executorch::ET_RUNTIME_NAMESPACE::Backend;
+using executorch::ET_RUNTIME_NAMESPACE::BackendExecutionContext;
+using executorch::ET_RUNTIME_NAMESPACE::BackendInitContext;
+using executorch::ET_RUNTIME_NAMESPACE::BackendInterface;
+using executorch::ET_RUNTIME_NAMESPACE::CompileSpec;
+using executorch::ET_RUNTIME_NAMESPACE::DelegateHandle;
 using executorch::runtime::ArrayRef;
-using executorch::runtime::Backend;
-using executorch::runtime::BackendExecutionContext;
-using executorch::runtime::BackendInitContext;
-using executorch::runtime::BackendInterface;
-using executorch::runtime::CompileSpec;
-using executorch::runtime::DelegateHandle;
 using executorch::runtime::Error;
 using executorch::runtime::EValue;
 using executorch::runtime::FreeableBuffer;
diff --git a/runtime/kernel/kernel_runtime_context.h b/runtime/kernel/kernel_runtime_context.h
index ad269f5dd4b..6facecc7632 100644
--- a/runtime/kernel/kernel_runtime_context.h
+++ b/runtime/kernel/kernel_runtime_context.h
@@ -15,7 +15,7 @@
 #include <executorch/runtime/platform/compiler.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
 /**
  * Runtime state and functionality for kernel implementations.
@@ -107,7 +107,7 @@ class KernelRuntimeContext {
   Error failure_state_ = Error::Ok;
 };
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 // TODO(T197294990): Remove these deprecated aliases once all users have moved
@@ -115,15 +115,15 @@ class KernelRuntimeContext {
 namespace torch {
 namespace executor {
 /// DEPRECATED: Use ::executorch::runtime::KernelRuntimeContext instead.
-using ::executorch::runtime::KernelRuntimeContext;
+using ::executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 /// DEPRECATED: Use ::executorch::runtime::KernelRuntimeContext instead.
-using RuntimeContext = ::executorch::runtime::KernelRuntimeContext;
+using RuntimeContext = ::executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 } // namespace executor
 } // namespace torch
 namespace executorch {
 namespace aten {
 /// DEPRECATED: Use ::executorch::runtime::KernelRuntimeContext instead.
-using RuntimeContext = ::executorch::runtime::KernelRuntimeContext;
+using RuntimeContext = ::executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 } // namespace aten
 } // namespace executorch
 // DEPRECATED: The executorch::aten:: namespace is deprecated. Use
diff --git a/runtime/kernel/operator_registry.cpp b/runtime/kernel/operator_registry.cpp
index 85705e5b3fd..d7e7b298c10 100644
--- a/runtime/kernel/operator_registry.cpp
+++ b/runtime/kernel/operator_registry.cpp
@@ -15,7 +15,7 @@
 #include <executorch/runtime/platform/system.h>
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
 namespace {
 
@@ -258,5 +258,5 @@ Span<const Kernel> get_registered_kernels() {
   return {registered_kernels, num_registered_kernels};
 }
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
diff --git a/runtime/kernel/operator_registry.h b/runtime/kernel/operator_registry.h
index a3cdcd66cee..f7a62208dd8 100644
--- a/runtime/kernel/operator_registry.h
+++ b/runtime/kernel/operator_registry.h
@@ -40,7 +40,7 @@
   }
 
 namespace executorch {
-namespace runtime {
+namespace ET_RUNTIME_NAMESPACE {
 
 class KernelRuntimeContext; // Forward declaration
 using OpFunction = void (*)(KernelRuntimeContext&, EValue**);
@@ -258,38 +258,41 @@ ET_NODISCARD inline Error register_kernel(const Kernel& kernel) {
   return register_kernels({&kernel, 1});
 };
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 namespace torch {
 namespace executor {
 // TODO(T197294990): Remove these deprecated aliases once all users have moved
 // to the new `::executorch` namespaces.
-using ::executorch::runtime::Kernel;
-using ::executorch::runtime::KernelKey;
-using ::executorch::runtime::KernelRuntimeContext;
-using ::executorch::runtime::OpFunction;
-using ::executorch::runtime::TensorMeta;
-using KernelRuntimeContext = ::executorch::runtime::KernelRuntimeContext;
+using ::executorch::ET_RUNTIME_NAMESPACE::Kernel;
+using ::executorch::ET_RUNTIME_NAMESPACE::KernelKey;
+using ::executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
+using ::executorch::ET_RUNTIME_NAMESPACE::OpFunction;
+using ::executorch::ET_RUNTIME_NAMESPACE::TensorMeta;
+using KernelRuntimeContext =
+    ::executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 
 inline ::executorch::runtime::Error register_kernels(ArrayRef<Kernel> kernels) {
-  return ::executorch::runtime::register_kernels(
+  return ::executorch::ET_RUNTIME_NAMESPACE::register_kernels(
       {kernels.data(), kernels.size()});
 }
 inline OpFunction getOpsFn(
     const char* name,
     ArrayRef<TensorMeta> meta_list = {}) {
-  auto result = ::executorch::runtime::get_op_function_from_registry(
-      name, {meta_list.data(), meta_list.size()});
+  auto result =
+      ::executorch::ET_RUNTIME_NAMESPACE::get_op_function_from_registry(
+          name, {meta_list.data(), meta_list.size()});
   ET_CHECK(result.ok()); // get_op_function_from_registry() logs details.
   return *result;
 }
 inline bool hasOpsFn(const char* name, ArrayRef<TensorMeta> meta_list = {}) {
-  return ::executorch::runtime::registry_has_op_function(
+  return ::executorch::ET_RUNTIME_NAMESPACE::registry_has_op_function(
       name, {meta_list.data(), meta_list.size()});
 }
 inline ArrayRef<Kernel> get_kernels() {
-  Span<const Kernel> kernels = ::executorch::runtime::get_registered_kernels();
+  Span<const Kernel> kernels =
+      ::executorch::ET_RUNTIME_NAMESPACE::get_registered_kernels();
   return ArrayRef<Kernel>(kernels.data(), kernels.size());
 }
 } // namespace executor
diff --git a/runtime/kernel/targets.bzl b/runtime/kernel/targets.bzl
index b6aa9d7a95e..8a945f19881 100644
--- a/runtime/kernel/targets.bzl
+++ b/runtime/kernel/targets.bzl
@@ -21,21 +21,6 @@ def define_common_targets():
     TARGETS and BUCK files that call this function.
     """
 
-    runtime.cxx_library(
-        name = "operator_registry",
-        srcs = ["operator_registry.cpp"],
-        exported_headers = ["operator_registry.h"],
-        visibility = [
-            "//executorch/...",
-            "@EXECUTORCH_CLIENTS",
-        ],
-        exported_deps = [
-            "//executorch/runtime/core:core",
-            "//executorch/runtime/core:evalue",
-        ],
-        preprocessor_flags = _operator_registry_preprocessor_flags(),
-    )
-
     runtime.cxx_library(
         name = "operator_registry_MAX_NUM_KERNELS_TEST_ONLY",
         srcs = ["operator_registry.cpp"],
@@ -68,6 +53,21 @@ def define_common_targets():
     for aten_mode in get_aten_mode_options():
         aten_suffix = "_aten" if aten_mode else ""
 
+        runtime.cxx_library(
+            name = "operator_registry" + aten_suffix,
+            srcs = ["operator_registry.cpp"],
+            exported_headers = ["operator_registry.h"],
+            visibility = [
+                "//executorch/...",
+                "@EXECUTORCH_CLIENTS",
+            ],
+            exported_deps = [
+                "//executorch/runtime/core:core",
+                "//executorch/runtime/core:evalue" + aten_suffix,
+            ],
+            preprocessor_flags = _operator_registry_preprocessor_flags(),
+        )
+
         runtime.cxx_library(
             name = "kernel_runtime_context" + aten_suffix,
             exported_headers = [
diff --git a/runtime/kernel/test/kernel_runtime_context_test.cpp b/runtime/kernel/test/kernel_runtime_context_test.cpp
index 50bd860fb9c..2c3b536b0d4 100644
--- a/runtime/kernel/test/kernel_runtime_context_test.cpp
+++ b/runtime/kernel/test/kernel_runtime_context_test.cpp
@@ -13,8 +13,8 @@
 #include <gtest/gtest.h>
 
 using namespace ::testing;
+using executorch::ET_RUNTIME_NAMESPACE::KernelRuntimeContext;
 using executorch::runtime::Error;
-using executorch::runtime::KernelRuntimeContext;
 using executorch::runtime::MemoryAllocator;
 using executorch::runtime::Result;
 
diff --git a/runtime/kernel/test/targets.bzl b/runtime/kernel/test/targets.bzl
index bd66fc05b6f..4b3ed0f3139 100644
--- a/runtime/kernel/test/targets.bzl
+++ b/runtime/kernel/test/targets.bzl
@@ -101,3 +101,16 @@ def define_common_targets():
                 ":specialized_kernel_generated_lib",
             ],
         )
+
+        if aten_mode:
+            # Make sure we can depend on both generated_lib and generated_lib_aten
+            # in the same binary.
+            runtime.cxx_test(
+                name = "test_generated_lib_and_aten",
+                srcs = ["test_generated_lib_and_aten.cpp"],
+                deps = [
+                    "//executorch/kernels/portable:generated_lib",
+                    "//executorch/kernels/portable:generated_lib_aten",
+                    "//executorch/runtime/kernel:operator_registry_aten",
+                ],
+            )
diff --git a/runtime/kernel/test/test_generated_lib_and_aten.cpp b/runtime/kernel/test/test_generated_lib_and_aten.cpp
new file mode 100644
index 00000000000..f9bfebc4a80
--- /dev/null
+++ b/runtime/kernel/test/test_generated_lib_and_aten.cpp
@@ -0,0 +1,45 @@
+/*
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
+ * All rights reserved.
+ *
+ * This source code is licensed under the BSD-style license found in the
+ * LICENSE file in the root directory of this source tree.
+ */
+
+#include <gtest/gtest.h>
+#include <vector>
+
+#include <executorch/runtime/core/exec_aten/exec_aten.h>
+#include <executorch/runtime/core/result.h>
+#include <executorch/runtime/core/span.h>
+#include <executorch/runtime/kernel/kernel_runtime_context.h>
+#include <executorch/runtime/kernel/operator_registry.h>
+#include <executorch/runtime/platform/runtime.h>
+#include <executorch/test/utils/DeathTest.h>
+
+using namespace ::testing;
+using executorch::aten::ScalarType;
+using executorch::runtime::Error;
+using executorch::runtime::EValue;
+
+class GeneratedLibAndAtenTest : public ::testing::Test {
+ public:
+  void SetUp() override {
+    executorch::runtime::runtime_init();
+  }
+};
+
+TEST_F(GeneratedLibAndAtenTest, GetKernelsFromATenRegistry) {
+  // Check if the kernel exists in the ATen registry
+  bool has_kernel =
+      executorch::runtime::aten::registry_has_op_function("aten::add.out");
+  EXPECT_TRUE(has_kernel)
+      << "Kernel 'aten::add.out' not found in ATen registry";
+
+  // Get the kernel from the ATen registry
+  auto result =
+      executorch::runtime::aten::get_op_function_from_registry("aten::add.out");
+  EXPECT_EQ(result.error(), Error::Ok)
+      << "Failed to get kernel from ATen registry";
+  EXPECT_NE(*result, nullptr) << "Kernel function from ATen registry is null";
+}
diff --git a/shim_et/xplat/executorch/codegen/codegen.bzl b/shim_et/xplat/executorch/codegen/codegen.bzl
index a6d6d59e0c2..e1cebaa1140 100644
--- a/shim_et/xplat/executorch/codegen/codegen.bzl
+++ b/shim_et/xplat/executorch/codegen/codegen.bzl
@@ -688,7 +688,7 @@ def executorch_generated_lib(
                 "ovr_config//os:windows": [],
             }) + compiler_flags,
             deps = [
-                "//executorch/runtime/kernel:operator_registry",
+                "//executorch/runtime/kernel:operator_registry" + aten_suffix,
                 "//executorch/kernels/prim_ops:prim_ops_registry" + aten_suffix,
                 "//executorch/runtime/core:evalue" + aten_suffix,
                 "//executorch/codegen:macros",