diff --git a/usaspending_api/common/spark/jobs.py b/usaspending_api/common/spark/jobs.py
index 9c48d6734a..fe397c2667 100644
--- a/usaspending_api/common/spark/jobs.py
+++ b/usaspending_api/common/spark/jobs.py
@@ -5,9 +5,12 @@
 from datetime import datetime
 from typing import TYPE_CHECKING, Generator
 
+import boto3
+from botocore.client import BaseClient
 from databricks.sdk import WorkspaceClient
 from databricks.sdk.config import Config as DatabricksConfig
 from databricks.sdk.service.jobs import BaseJob, RunLifeCycleState
+from django.conf import settings
 from django.core.management import call_command
 from duckdb.experimental.spark.sql import SparkSession as DuckDBSparkSession
 
@@ -109,13 +112,65 @@ def _wait_for_run_to_start(self, job: BaseJob, job_run_id: int) -> None:
 
 
 class EmrServerlessStrategy(_AbstractStrategy):
+    _client: BaseClient = None
+
     @property
     def name(self) -> str:
         return "EMR_SERVERLESS"
 
+    @property
+    def client(self) -> BaseClient:
+        if not self._client:
+            self._client = boto3.client("emr-serverless", settings.USASPENDING_AWS_REGION)
+        return self._client
+
+    def _get_application_id(self, application_name: str) -> str:
+        paginator = self.client.get_paginator("list_applications")
+        matched_applications = []
+        for list_applications_response in paginator.paginate():
+            temp_applications = list_applications_response.get("applications", [])
+            matched_applications.extend(
+                [application for application in temp_applications if application["name"] == application_name]
+            )
+
+        match len(matched_applications):
+            case 1:
+                application_id = matched_applications[0]["id"]
+            case 0:
+                raise ValueError(f"No EMR Serverless application found with name '{application_name}'")
+            case _:
+                arns_to_log = [application["arn"] for application in matched_applications]
+                raise ValueError(
+                    f"More than 1 EMR Serverless application found with name '{application_name}': {arns_to_log}"
+                )
+
+        return application_id
+
     def handle_start(self, job_name: str, command_name: str, command_options: list[str], **kwargs) -> dict:
-        # TODO: This will be implemented as we migrate, but added as a placeholder for now
-        pass
+        application_id = kwargs.get("application_id")
+        application_name = kwargs.get("application_name")
+        execution_role_arn = kwargs.get("execution_role_arn")
+
+        if not execution_role_arn:
+            raise ValueError(f"Execution role ARN is required to start an EMR Serverless job")
+        elif not application_name and not application_id:
+            raise ValueError(f"Application Name or ID is required to start an EMR Serverless job")
+        elif application_name and not application_id:
+            application_id = self._get_application_id(application_name)
+
+        response = self.client.start_job_run(
+            applicationId=application_id,
+            executionRoleArn=execution_role_arn,
+            name=job_name,
+            mode="BATCH",
+            jobDriver={
+                "sparkSubmit": {
+                    "entryPoint": command_name,
+                    "entryPointArguments": command_options,
+                }
+            },
+        )
+        return response
 
 
 class LocalStrategy(_AbstractStrategy):
diff --git a/usaspending_api/common/tests/unit/test_spark_jobs.py b/usaspending_api/common/tests/unit/test_spark_jobs.py
index 6a3e769a04..8bc5eb4eb8 100644
--- a/usaspending_api/common/tests/unit/test_spark_jobs.py
+++ b/usaspending_api/common/tests/unit/test_spark_jobs.py
@@ -51,3 +51,49 @@ def test_databricks_strategy_handle_start(databricks_strategy_client):
 
     spark_job = SparkJobs(DatabricksStrategy())
     assert spark_job.start(job_name="", command_name="", command_options=[""]) == {"job_id": 1, "run_id": 10}
+
+
+@patch("usaspending_api.common.spark.jobs.EmrServerlessStrategy.client")
+def test_emr_serverless_strategy_handle_start(emr_serverless_strategy_client):
+    mock_application = MagicMock()
+    mock_application.application_id = 1
+    mock_application.name = "application_1"
+
+    mock_application_paginator = MagicMock()
+    mock_application_paginator.paginate = MagicMock(
+        return_value=[{"applications": [{"id": mock_application.application_id, "name": mock_application.name}]}]
+    )
+
+    emr_serverless_strategy_client.get_paginator = MagicMock(return_value=mock_application_paginator)
+    emr_serverless_strategy_client.start_job_run = MagicMock()
+
+    strategy = EmrServerlessStrategy()
+    assert strategy._get_application_id("application_1") == 1
+
+    emr_serverless_strategy_client.reset_mock()
+
+    spark_job = SparkJobs(strategy)
+    spark_job.start(
+        job_name="",
+        command_name="",
+        command_options=[""],
+        application_id="Some ID",
+        application_name="application_1",
+        execution_role_arn="Some ARN",
+    )
+    assert emr_serverless_strategy_client.get_paginator.call_count == 0
+    assert emr_serverless_strategy_client.start_job_run.call_count == 1
+
+    emr_serverless_strategy_client.reset_mock()
+
+    spark_job = SparkJobs(strategy)
+    spark_job.start(
+        job_name="",
+        command_name="",
+        command_options=[""],
+        application_id=None,
+        application_name="application_1",
+        execution_role_arn="Some ARN",
+    )
+    assert emr_serverless_strategy_client.get_paginator.call_count == 1
+    assert emr_serverless_strategy_client.start_job_run.call_count == 1