diff --git a/backends/cadence/aot/TARGETS b/backends/cadence/aot/TARGETS
index ae60c299f2c..ff4256d19f6 100644
--- a/backends/cadence/aot/TARGETS
+++ b/backends/cadence/aot/TARGETS
@@ -49,7 +49,7 @@ python_library(
 python_library(
     name = "passes",
     srcs = [
-        "passes.py",
+        "_passes.py",
     ],
     deps = [
         ":utils",
diff --git a/backends/cadence/aot/passes.py b/backends/cadence/aot/_passes.py
similarity index 100%
rename from backends/cadence/aot/passes.py
rename to backends/cadence/aot/_passes.py
diff --git a/backends/cadence/aot/compiler.py b/backends/cadence/aot/compiler.py
index 5b151a3b6a4..89467dbc234 100644
--- a/backends/cadence/aot/compiler.py
+++ b/backends/cadence/aot/compiler.py
@@ -11,7 +11,7 @@
 
 import torch
 
-from executorch.backends.cadence.aot.passes import (
+from executorch.backends.cadence.aot._passes import (
     InitializePipeline,
     RemoveNopExpandOpPass,
     RemoveZeroSizedCatArgsPass,
diff --git a/backends/mediatek/passes/__init__.py b/backends/mediatek/_passes/__init__.py
similarity index 100%
rename from backends/mediatek/passes/__init__.py
rename to backends/mediatek/_passes/__init__.py
diff --git a/backends/mediatek/passes/decompose_scaled_dot_product_attention.py b/backends/mediatek/_passes/decompose_scaled_dot_product_attention.py
similarity index 100%
rename from backends/mediatek/passes/decompose_scaled_dot_product_attention.py
rename to backends/mediatek/_passes/decompose_scaled_dot_product_attention.py
diff --git a/backends/mediatek/quantizer/quantizer.py b/backends/mediatek/quantizer/quantizer.py
index 44e35ef54eb..4e78d6dff1a 100644
--- a/backends/mediatek/quantizer/quantizer.py
+++ b/backends/mediatek/quantizer/quantizer.py
@@ -7,7 +7,7 @@
 from torch.ao.quantization.quantizer import Quantizer
 from torch.fx import GraphModule
 
-from ..passes.decompose_scaled_dot_product_attention import (
+from .._passes.decompose_scaled_dot_product_attention import (
     DecomposeScaledDotProductAttention,
 )
 from .annotator import annotate
diff --git a/backends/qualcomm/README.md b/backends/qualcomm/README.md
index be7cd427d6e..d426aff16b1 100644
--- a/backends/qualcomm/README.md
+++ b/backends/qualcomm/README.md
@@ -46,7 +46,7 @@ backends/qualcomm
 |   └── python # Python interface for using QNN libraries.
 ├── builders # Codes for lowering each operators (AoT Part).
 ├── partition # QNN Partitioner (AoT Part).
-├── passes # Various passes helping lower models to QNN backend (AoT Part).
+├── _passes # Various private passes helping lower models to QNN backend (AoT Part).
 ├── python # Places to put pybind artifacts for accessing QNN APIs, structures, etc (AoT Part).
 ├── quantizer # QNN Quantizer
 ├── runtime # Here is QNN runtime responsbile for compiling a model on x64.
diff --git a/backends/qualcomm/TARGETS b/backends/qualcomm/TARGETS
index 6a720a62479..895a5e8aae4 100644
--- a/backends/qualcomm/TARGETS
+++ b/backends/qualcomm/TARGETS
@@ -13,6 +13,6 @@ runtime.python_library(
         "@EXECUTORCH_CLIENTS",
     ],
     deps = [
-        "//executorch/backends/qualcomm/passes:passes",
+        "//executorch/backends/qualcomm/_passes:passes",
     ],
 )
diff --git a/backends/qualcomm/passes/TARGETS b/backends/qualcomm/_passes/TARGETS
similarity index 100%
rename from backends/qualcomm/passes/TARGETS
rename to backends/qualcomm/_passes/TARGETS
diff --git a/backends/qualcomm/passes/annotate_and_quant_scalar.py b/backends/qualcomm/_passes/annotate_and_quant_scalar.py
similarity index 100%
rename from backends/qualcomm/passes/annotate_and_quant_scalar.py
rename to backends/qualcomm/_passes/annotate_and_quant_scalar.py
diff --git a/backends/qualcomm/passes/annotate_decomposed.py b/backends/qualcomm/_passes/annotate_decomposed.py
similarity index 100%
rename from backends/qualcomm/passes/annotate_decomposed.py
rename to backends/qualcomm/_passes/annotate_decomposed.py
diff --git a/backends/qualcomm/passes/annotate_quant_attrs.py b/backends/qualcomm/_passes/annotate_quant_attrs.py
similarity index 100%
rename from backends/qualcomm/passes/annotate_quant_attrs.py
rename to backends/qualcomm/_passes/annotate_quant_attrs.py
diff --git a/backends/qualcomm/passes/build_quant_io.py b/backends/qualcomm/_passes/build_quant_io.py
similarity index 100%
rename from backends/qualcomm/passes/build_quant_io.py
rename to backends/qualcomm/_passes/build_quant_io.py
diff --git a/backends/qualcomm/passes/convert_binary_op_with_scalar.py b/backends/qualcomm/_passes/convert_binary_op_with_scalar.py
similarity index 100%
rename from backends/qualcomm/passes/convert_binary_op_with_scalar.py
rename to backends/qualcomm/_passes/convert_binary_op_with_scalar.py
diff --git a/backends/qualcomm/passes/convert_bmm_to_matmul.py b/backends/qualcomm/_passes/convert_bmm_to_matmul.py
similarity index 100%
rename from backends/qualcomm/passes/convert_bmm_to_matmul.py
rename to backends/qualcomm/_passes/convert_bmm_to_matmul.py
diff --git a/backends/qualcomm/passes/convert_interpolate_with_upsample2d.py b/backends/qualcomm/_passes/convert_interpolate_with_upsample2d.py
similarity index 100%
rename from backends/qualcomm/passes/convert_interpolate_with_upsample2d.py
rename to backends/qualcomm/_passes/convert_interpolate_with_upsample2d.py
diff --git a/backends/qualcomm/passes/convert_prelu.py b/backends/qualcomm/_passes/convert_prelu.py
similarity index 100%
rename from backends/qualcomm/passes/convert_prelu.py
rename to backends/qualcomm/_passes/convert_prelu.py
diff --git a/backends/qualcomm/passes/convert_to_linear.py b/backends/qualcomm/_passes/convert_to_linear.py
similarity index 100%
rename from backends/qualcomm/passes/convert_to_linear.py
rename to backends/qualcomm/_passes/convert_to_linear.py
diff --git a/backends/qualcomm/passes/decompose_silu.py b/backends/qualcomm/_passes/decompose_silu.py
similarity index 100%
rename from backends/qualcomm/passes/decompose_silu.py
rename to backends/qualcomm/_passes/decompose_silu.py
diff --git a/backends/qualcomm/passes/fold_qdq.py b/backends/qualcomm/_passes/fold_qdq.py
similarity index 100%
rename from backends/qualcomm/passes/fold_qdq.py
rename to backends/qualcomm/_passes/fold_qdq.py
diff --git a/backends/qualcomm/passes/fuse_consecutive_transpose.py b/backends/qualcomm/_passes/fuse_consecutive_transpose.py
similarity index 100%
rename from backends/qualcomm/passes/fuse_consecutive_transpose.py
rename to backends/qualcomm/_passes/fuse_consecutive_transpose.py
diff --git a/backends/qualcomm/passes/i64_to_i32.py b/backends/qualcomm/_passes/i64_to_i32.py
similarity index 100%
rename from backends/qualcomm/passes/i64_to_i32.py
rename to backends/qualcomm/_passes/i64_to_i32.py
diff --git a/backends/qualcomm/passes/insert_io_qdq.py b/backends/qualcomm/_passes/insert_io_qdq.py
similarity index 100%
rename from backends/qualcomm/passes/insert_io_qdq.py
rename to backends/qualcomm/_passes/insert_io_qdq.py
diff --git a/backends/qualcomm/passes/insert_requantize.py b/backends/qualcomm/_passes/insert_requantize.py
similarity index 100%
rename from backends/qualcomm/passes/insert_requantize.py
rename to backends/qualcomm/_passes/insert_requantize.py
diff --git a/backends/qualcomm/passes/layout_transform.py b/backends/qualcomm/_passes/layout_transform.py
similarity index 100%
rename from backends/qualcomm/passes/layout_transform.py
rename to backends/qualcomm/_passes/layout_transform.py
diff --git a/backends/qualcomm/passes/recompose_pixel_unshuffle.py b/backends/qualcomm/_passes/recompose_pixel_unshuffle.py
similarity index 100%
rename from backends/qualcomm/passes/recompose_pixel_unshuffle.py
rename to backends/qualcomm/_passes/recompose_pixel_unshuffle.py
diff --git a/backends/qualcomm/passes/recompose_rms_norm.py b/backends/qualcomm/_passes/recompose_rms_norm.py
similarity index 100%
rename from backends/qualcomm/passes/recompose_rms_norm.py
rename to backends/qualcomm/_passes/recompose_rms_norm.py
diff --git a/backends/qualcomm/passes/reduce_dynamic_range.py b/backends/qualcomm/_passes/reduce_dynamic_range.py
similarity index 100%
rename from backends/qualcomm/passes/reduce_dynamic_range.py
rename to backends/qualcomm/_passes/reduce_dynamic_range.py
diff --git a/backends/qualcomm/passes/remove_redundancy.py b/backends/qualcomm/_passes/remove_redundancy.py
similarity index 100%
rename from backends/qualcomm/passes/remove_redundancy.py
rename to backends/qualcomm/_passes/remove_redundancy.py
diff --git a/backends/qualcomm/passes/replace_index_put_input.py b/backends/qualcomm/_passes/replace_index_put_input.py
similarity index 100%
rename from backends/qualcomm/passes/replace_index_put_input.py
rename to backends/qualcomm/_passes/replace_index_put_input.py
diff --git a/backends/qualcomm/passes/replace_inf_buffer.py b/backends/qualcomm/_passes/replace_inf_buffer.py
similarity index 100%
rename from backends/qualcomm/passes/replace_inf_buffer.py
rename to backends/qualcomm/_passes/replace_inf_buffer.py
diff --git a/backends/qualcomm/passes/utils.py b/backends/qualcomm/_passes/utils.py
similarity index 100%
rename from backends/qualcomm/passes/utils.py
rename to backends/qualcomm/_passes/utils.py
diff --git a/backends/qualcomm/qnn_preprocess.py b/backends/qualcomm/qnn_preprocess.py
index 6360b35280f..417934acbd4 100644
--- a/backends/qualcomm/qnn_preprocess.py
+++ b/backends/qualcomm/qnn_preprocess.py
@@ -11,15 +11,15 @@
 import executorch.backends.qualcomm.python.PyQnnManagerAdaptor as PyQnnManager
 
 import torch  # noqa: F401
-from executorch.backends.qualcomm.builders.node_visitor import get_node_visitors
-from executorch.backends.qualcomm.builders.qnn_constants import OpContextLoader
-from executorch.backends.qualcomm.passes.convert_to_linear import ConvertToLinear
-from executorch.backends.qualcomm.passes.fuse_consecutive_transpose import (
+from executorch.backends.qualcomm._passes.convert_to_linear import ConvertToLinear
+from executorch.backends.qualcomm._passes.fuse_consecutive_transpose import (
     FuseConsecutiveTranspose,
 )
-from executorch.backends.qualcomm.passes.insert_io_qdq import InsertIOQDQ
-from executorch.backends.qualcomm.passes.insert_requantize import InsertRequantize
-from executorch.backends.qualcomm.passes.layout_transform import LayoutTransform
+from executorch.backends.qualcomm._passes.insert_io_qdq import InsertIOQDQ
+from executorch.backends.qualcomm._passes.insert_requantize import InsertRequantize
+from executorch.backends.qualcomm._passes.layout_transform import LayoutTransform
+from executorch.backends.qualcomm.builders.node_visitor import get_node_visitors
+from executorch.backends.qualcomm.builders.qnn_constants import OpContextLoader
 from executorch.backends.qualcomm.utils.utils import generate_qnn_executorch_option
 from executorch.exir.backend.backend_details import (
     BackendDetails,
diff --git a/backends/qualcomm/quantizer/quantizer.py b/backends/qualcomm/quantizer/quantizer.py
index 1d4b4c2f217..ce8fab5e6f2 100644
--- a/backends/qualcomm/quantizer/quantizer.py
+++ b/backends/qualcomm/quantizer/quantizer.py
@@ -7,12 +7,12 @@
 from typing import Callable, Dict, Optional, Sequence, Set
 
 import torch
-from executorch.backends.qualcomm.passes.decompose_silu import DecomposeSilu
-from executorch.backends.qualcomm.passes.recompose_pixel_unshuffle import (
+from executorch.backends.qualcomm._passes.decompose_silu import DecomposeSilu
+from executorch.backends.qualcomm._passes.recompose_pixel_unshuffle import (
     RecomposePixelUnshuffle,
 )
-from executorch.backends.qualcomm.passes.reduce_dynamic_range import ReduceDynamicRange
-from executorch.backends.qualcomm.passes.replace_inf_buffer import ReplaceInfBuffer
+from executorch.backends.qualcomm._passes.reduce_dynamic_range import ReduceDynamicRange
+from executorch.backends.qualcomm._passes.replace_inf_buffer import ReplaceInfBuffer
 from executorch.backends.transforms.decompose_sdpa import (
     DecomposeScaledDotProductAttention,
 )
diff --git a/backends/qualcomm/utils/constants.py b/backends/qualcomm/utils/constants.py
index 7a0be8b0703..8a37b2bd8ca 100644
--- a/backends/qualcomm/utils/constants.py
+++ b/backends/qualcomm/utils/constants.py
@@ -6,7 +6,7 @@
 
 # Qualcomm specific key
 
-# constants in backends/qualcomm/passes & backends/qualcomm/builders
+# constants in backends/qualcomm/_passes & backends/qualcomm/builders
 QCOM_AXIS = "axis"
 QCOM_AXIS_ORDER = "axis_order"
 QCOM_BITWIDTH = "bitwidth"
diff --git a/backends/qualcomm/utils/utils.py b/backends/qualcomm/utils/utils.py
index 56f71a722a0..3c93f75ec55 100644
--- a/backends/qualcomm/utils/utils.py
+++ b/backends/qualcomm/utils/utils.py
@@ -14,40 +14,42 @@
 import executorch.exir as exir
 
 import torch
-
-from executorch.backends.qualcomm.builders.node_visitor import (
-    QNN_QUANT_TYPE_MAP,
-    QNN_TENSOR_TYPE_MAP,
-)
-from executorch.backends.qualcomm.builders.qnn_constants import OpContextLoader
-from executorch.backends.qualcomm.passes.annotate_and_quant_scalar import (
+from executorch.backends.qualcomm._passes.annotate_and_quant_scalar import (
     AnnotateAndQuantScalar,
 )
-from executorch.backends.qualcomm.passes.annotate_decomposed import AnnotateDecomposed
-from executorch.backends.qualcomm.passes.annotate_quant_attrs import AnnotateQuantAttrs
-from executorch.backends.qualcomm.passes.convert_binary_op_with_scalar import (
+from executorch.backends.qualcomm._passes.annotate_decomposed import AnnotateDecomposed
+from executorch.backends.qualcomm._passes.annotate_quant_attrs import AnnotateQuantAttrs
+from executorch.backends.qualcomm._passes.convert_binary_op_with_scalar import (
     ConvertBinaryOpsWithScalar,
 )
-from executorch.backends.qualcomm.passes.convert_bmm_to_matmul import ConvertBmmToMatmul
-from executorch.backends.qualcomm.passes.convert_interpolate_with_upsample2d import (
-    ConvertInterpolateWithUpsample2D,
+from executorch.backends.qualcomm._passes.convert_bmm_to_matmul import (
+    ConvertBmmToMatmul,
 )
-from executorch.backends.qualcomm.passes.convert_prelu import ConvertPReLU
-from executorch.backends.qualcomm.passes.convert_to_linear import ConvertToLinear
-from executorch.backends.qualcomm.passes.expand_broadcast_tensor_shape import (
-    ExpandBroadcastTensorShape,
+from executorch.backends.qualcomm._passes.convert_interpolate_with_upsample2d import (
+    ConvertInterpolateWithUpsample2D,
 )
-from executorch.backends.qualcomm.passes.fold_qdq import FoldQDQ
-from executorch.backends.qualcomm.passes.i64_to_i32 import I64toI32
-from executorch.backends.qualcomm.passes.layout_transform import LayoutTransform
-from executorch.backends.qualcomm.passes.recompose_pixel_unshuffle import (
+from executorch.backends.qualcomm._passes.convert_prelu import ConvertPReLU
+from executorch.backends.qualcomm._passes.convert_to_linear import ConvertToLinear
+from executorch.backends.qualcomm._passes.fold_qdq import FoldQDQ
+from executorch.backends.qualcomm._passes.i64_to_i32 import I64toI32
+from executorch.backends.qualcomm._passes.layout_transform import LayoutTransform
+from executorch.backends.qualcomm._passes.recompose_pixel_unshuffle import (
     RecomposePixelUnshuffle,
 )
-from executorch.backends.qualcomm.passes.recompose_rms_norm import RecomposeRmsNorm
-from executorch.backends.qualcomm.passes.remove_redundancy import RemoveRedundancy
-from executorch.backends.qualcomm.passes.replace_index_put_input import (
+from executorch.backends.qualcomm._passes.recompose_rms_norm import RecomposeRmsNorm
+from executorch.backends.qualcomm._passes.remove_redundancy import RemoveRedundancy
+from executorch.backends.qualcomm._passes.replace_index_put_input import (
     ReplaceIndexPutInput,
 )
+
+from executorch.backends.qualcomm.builders.node_visitor import (
+    QNN_QUANT_TYPE_MAP,
+    QNN_TENSOR_TYPE_MAP,
+)
+from executorch.backends.qualcomm.builders.qnn_constants import OpContextLoader
+from executorch.backends.qualcomm.passes.expand_broadcast_tensor_shape import (
+    ExpandBroadcastTensorShape,
+)
 from executorch.backends.qualcomm.serialization.qnn_compile_spec_schema import (
     _soc_info_table,
     QcomChipset,
diff --git a/backends/transforms/fuse_conv_with_clamp.py b/backends/transforms/fuse_conv_with_clamp.py
index 3f18402ee35..15973cae577 100644
--- a/backends/transforms/fuse_conv_with_clamp.py
+++ b/backends/transforms/fuse_conv_with_clamp.py
@@ -7,7 +7,9 @@
 import sys
 
 import torch
-from executorch.backends.vulkan.passes.custom_ops_defs import conv_with_clamp_op  # noqa
+from executorch.backends.vulkan._passes.custom_ops_defs import (  # noqa
+    conv_with_clamp_op,
+)
 
 from executorch.exir.dialects._ops import ops as exir_ops
 from executorch.exir.pass_base import ExportPass, PassResult
diff --git a/backends/transforms/targets.bzl b/backends/transforms/targets.bzl
index 458a2d71bb5..47c518a8637 100644
--- a/backends/transforms/targets.bzl
+++ b/backends/transforms/targets.bzl
@@ -70,7 +70,7 @@ def define_common_targets():
         deps = [
             ":utils",
             "//caffe2:torch",
-            "//executorch/backends/vulkan/passes:custom_ops_defs",
+            "//executorch/backends/vulkan/_passes:custom_ops_defs",
             "//executorch/exir:pass_base",
             "//executorch/exir:sym_util",
             "//executorch/exir/dialects:lib",
diff --git a/backends/vulkan/TARGETS b/backends/vulkan/TARGETS
index 8932d08b76f..bbddb70735c 100644
--- a/backends/vulkan/TARGETS
+++ b/backends/vulkan/TARGETS
@@ -28,7 +28,7 @@ runtime.python_library(
         "//executorch/backends/transforms:fuse_view_copy",
         "//executorch/backends/transforms:mean_to_sum_div",
         "//executorch/backends/transforms:remove_clone_ops",
-        "//executorch/backends/vulkan/passes:remove_local_scalar_dense",
+        "//executorch/backends/vulkan/_passes:remove_local_scalar_dense",
         "//executorch/exir:graph_module",
         "//executorch/exir/_serialize:_bindings",
         "//executorch/exir/_serialize:lib",
diff --git a/backends/vulkan/passes/TARGETS b/backends/vulkan/_passes/TARGETS
similarity index 100%
rename from backends/vulkan/passes/TARGETS
rename to backends/vulkan/_passes/TARGETS
diff --git a/backends/vulkan/passes/custom_ops_defs.py b/backends/vulkan/_passes/custom_ops_defs.py
similarity index 100%
rename from backends/vulkan/passes/custom_ops_defs.py
rename to backends/vulkan/_passes/custom_ops_defs.py
diff --git a/backends/vulkan/passes/remove_local_scalar_dense_ops.py b/backends/vulkan/_passes/remove_local_scalar_dense_ops.py
similarity index 100%
rename from backends/vulkan/passes/remove_local_scalar_dense_ops.py
rename to backends/vulkan/_passes/remove_local_scalar_dense_ops.py
diff --git a/backends/vulkan/passes/test_custom_ops.py b/backends/vulkan/_passes/test_custom_ops.py
similarity index 100%
rename from backends/vulkan/passes/test_custom_ops.py
rename to backends/vulkan/_passes/test_custom_ops.py
diff --git a/backends/vulkan/partitioner/supported_ops.py b/backends/vulkan/partitioner/supported_ops.py
index e9ed1439cd2..587485a5c99 100644
--- a/backends/vulkan/partitioner/supported_ops.py
+++ b/backends/vulkan/partitioner/supported_ops.py
@@ -8,7 +8,7 @@
 
 import operator
 
-from executorch.backends.vulkan.passes.custom_ops_defs import (  # noqa
+from executorch.backends.vulkan._passes.custom_ops_defs import (  # noqa
     conv_with_clamp_op,
     grid_priors_op,
 )
diff --git a/backends/vulkan/vulkan_preprocess.py b/backends/vulkan/vulkan_preprocess.py
index 18b5ad872f5..8c8d5ad27f9 100644
--- a/backends/vulkan/vulkan_preprocess.py
+++ b/backends/vulkan/vulkan_preprocess.py
@@ -17,7 +17,7 @@
 from executorch.backends.transforms.mean_to_sum_div import MeanToSumDiv
 from executorch.backends.transforms.remove_clone_ops import RemoveCloneOpsTransform
 
-from executorch.backends.vulkan.passes.remove_local_scalar_dense_ops import (
+from executorch.backends.vulkan._passes.remove_local_scalar_dense_ops import (
     RemoveLocalScalarDenseOpsTransform,
 )
 
diff --git a/examples/qualcomm/oss_scripts/llama2/llama.py b/examples/qualcomm/oss_scripts/llama2/llama.py
index 222f2717ed2..3dc1289ddb5 100644
--- a/examples/qualcomm/oss_scripts/llama2/llama.py
+++ b/examples/qualcomm/oss_scripts/llama2/llama.py
@@ -12,9 +12,9 @@
 from multiprocessing.connection import Client
 
 import torch
+from executorch.backends.qualcomm._passes.build_quant_io import BuildQuantIo
 
 from executorch.backends.qualcomm.partition.qnn_partitioner import QnnPartitioner
-from executorch.backends.qualcomm.passes.build_quant_io import BuildQuantIo
 
 from executorch.backends.qualcomm.quantizer.quantizer import QuantDtype
 from executorch.backends.qualcomm.serialization.qnn_compile_spec_schema import (