diff --git a/model-engine/model_engine_server/api/app.py b/model-engine/model_engine_server/api/app.py
index 851f01833..33e6830a8 100644
--- a/model-engine/model_engine_server/api/app.py
+++ b/model-engine/model_engine_server/api/app.py
@@ -33,7 +33,7 @@
 
 logger = make_logger(logger_name())
 
-# Allows us to make the Uvicorn worker concurrency in model_engine_server/api/worker.py very high
+# See also Uvicorn worker concurrency in model_engine_server/api/worker.py
 MAX_CONCURRENCY = 500
 
 concurrency_limiter = MultiprocessingConcurrencyLimiter(
diff --git a/model-engine/model_engine_server/api/worker.py b/model-engine/model_engine_server/api/worker.py
deleted file mode 100644
index 289640c88..000000000
--- a/model-engine/model_engine_server/api/worker.py
+++ /dev/null
@@ -1,14 +0,0 @@
-from uvicorn.workers import UvicornWorker
-
-# Gunicorn returns 503 instead of 429 when concurrency exceeds the limit
-# We'll autoscale at target concurrency of a much lower number (around 50), and this just makes sure we don't 503 with bursty traffic
-# We set this very high since model_engine_server/api/app.py sets a lower per-pod concurrency at which we start returning 429s
-CONCURRENCY_LIMIT = 10000
-
-
-class LaunchWorker(UvicornWorker):
-    """Overrides the configuration of the Uvicorn Worker."""
-
-    # uvloop and httptools are both faster than their alternatives, but they are not compatible
-    # with Windows or PyPy.
-    CONFIG_KWARGS = {"loop": "uvloop", "http": "httptools", "limit_concurrency": CONCURRENCY_LIMIT}
diff --git a/model-engine/model_engine_server/entrypoints/start_fastapi_server.py b/model-engine/model_engine_server/entrypoints/start_fastapi_server.py
index 119935ffb..eed6a88ae 100644
--- a/model-engine/model_engine_server/entrypoints/start_fastapi_server.py
+++ b/model-engine/model_engine_server/entrypoints/start_fastapi_server.py
@@ -3,28 +3,42 @@
 
 You can do this with `start-fastapi-server`.
 """
+
 import argparse
 import subprocess
 from typing import List
 
+# Uvicorn returns 503 instead of 429 when concurrency exceeds the limit
+# We'll autoscale at target concurrency of a much lower number (around 50), and this just makes sure we don't 503 with bursty traffic
+# We set this very high since model_engine_server/api/app.py sets a lower per-pod concurrency at which we start returning 429s
+CONCURRENCY_LIMIT = 10000
+
 
-def start_gunicorn_server(port: int, num_workers: int, debug: bool) -> None:
-    """Starts a GUnicorn server locally."""
+def start_uvicorn_server(port: int, debug: bool) -> None:
+    """Starts a Uvicorn server locally."""
     additional_args: List[str] = []
     if debug:
-        additional_args.extend(["--reload", "--timeout", "0"])
+        additional_args.extend(["--reload", "--timeout-graceful-shutdown", "0"])
     command = [
-        "gunicorn",
-        "--bind",
-        f"[::]:{port}",
-        "--timeout",
+        "uvicorn",
+        "--host",
+        "::",
+        "--port",
+        f"{port}",
+        "--timeout-graceful-shutdown",
         "60",
-        "--keep-alive",
+        "--timeout-keep-alive",
         "2",
-        "--worker-class",
-        "model_engine_server.api.worker.LaunchWorker",
+        # uvloop and httptools are both faster than their alternatives, but they are not compatible
+        # with Windows or PyPy.
+        "--loop",
+        "uvloop",
+        "--http",
+        "httptools",
+        "--limit-concurrency",
+        f"{CONCURRENCY_LIMIT}",
         "--workers",
-        f"{num_workers}",
+        "1",  # Let the Kubernetes deployment handle the number of pods
         *additional_args,
         "model_engine_server.api.app:app",
     ]
@@ -38,11 +52,10 @@ def entrypoint():
     # We can probably use asyncio since this service is going to be more I/O bound.
     parser = argparse.ArgumentParser(description="Hosted Inference Server")
     parser.add_argument("--port", type=int, default=5000)
-    parser.add_argument("--num-workers", type=int, default=4)
     parser.add_argument("--debug", "-d", action="store_true")
     args = parser.parse_args()
 
-    start_gunicorn_server(args.port, args.num_workers, args.debug)
+    start_uvicorn_server(args.port, args.debug)
 
 
 if __name__ == "__main__":
diff --git a/model-engine/requirements.in b/model-engine/requirements.in
index 2ef63150f..0a6e70e75 100644
--- a/model-engine/requirements.in
+++ b/model-engine/requirements.in
@@ -57,6 +57,6 @@ tokenizers~=0.15.2
 tqdm~=4.64
 transformers==4.38.0
 twine==3.7.1
-uvicorn==0.17.6
+uvicorn==0.30.0
 uvloop==0.17.0
 yarl~=1.4
\ No newline at end of file
diff --git a/model-engine/requirements.txt b/model-engine/requirements.txt
index 71e7440d8..546844085 100644
--- a/model-engine/requirements.txt
+++ b/model-engine/requirements.txt
@@ -23,8 +23,6 @@ anyio==3.7.1
     #   azure-core
     #   httpx
     #   starlette
-asgiref==3.7.2
-    # via uvicorn
 asn1crypto==1.5.1
     # via scramp
 async-timeout==4.0.2
@@ -76,9 +74,7 @@ boto3==1.28.1
     #   celery
     #   kombu
 boto3-stubs[essential]==1.26.67
-    # via
-    #   -r model-engine/requirements.in
-    #   boto3-stubs
+    # via -r model-engine/requirements.in
 botocore==1.31.1
     # via
     #   -r model-engine/requirements.in
@@ -95,9 +91,7 @@ cachetools==5.3.1
 cattrs==23.1.2
     # via ddtrace
 celery[redis,sqs,tblib]==5.3.6
-    # via
-    #   -r model-engine/requirements.in
-    #   celery
+    # via -r model-engine/requirements.in
 certifi==2023.7.22
     # via
     #   datadog-api-client
@@ -140,7 +134,6 @@ cryptography==42.0.5
     #   azure-storage-blob
     #   msal
     #   pyjwt
-    #   secretstorage
 dataclasses-json==0.5.9
     # via -r model-engine/requirements.in
 datadog==0.47.0
@@ -159,7 +152,7 @@ docutils==0.20.1
     # via readme-renderer
 envier==0.4.0
     # via ddtrace
-exceptiongroup==1.2.0
+exceptiongroup==1.2.1
     # via
     #   anyio
     #   cattrs
@@ -224,7 +217,7 @@ importlib-metadata==6.8.0
     #   keyring
     #   quart
     #   twine
-importlib-resources==6.1.1
+importlib-resources==6.4.0
     # via
     #   alembic
     #   jsonschema
@@ -242,10 +235,6 @@ itsdangerous==2.1.2
     #   starlette
 jaraco-classes==3.3.0
     # via keyring
-jeepney==0.8.0
-    # via
-    #   keyring
-    #   secretstorage
 jinja2==3.0.3
     # via
     #   -r model-engine/requirements.in
@@ -371,9 +360,7 @@ pygments==2.15.1
     #   readme-renderer
     #   rich
 pyjwt[crypto]==2.8.0
-    # via
-    #   msal
-    #   pyjwt
+    # via msal
 pyproject-hooks==1.0.0
     # via build
 python-dateutil==2.8.2
@@ -445,8 +432,6 @@ safetensors==0.4.2
     # via transformers
 scramp==1.4.4
     # via pg8000
-secretstorage==3.3.3
-    # via keyring
 sentencepiece==0.1.99
     # via -r model-engine/requirements.in
 sh==1.14.3
@@ -479,7 +464,6 @@ sqlalchemy[asyncio]==2.0.4
     # via
     #   -r model-engine/requirements.in
     #   alembic
-    #   sqlalchemy
 sse-starlette==1.6.1
     # via -r model-engine/requirements.in
 sseclient-py==1.7.2
@@ -489,7 +473,6 @@ starlette[full]==0.36.3
     #   -r model-engine/requirements.in
     #   fastapi
     #   sse-starlette
-    #   starlette
 stringcase==1.2.0
     # via -r model-engine/requirements.in
 tblib==2.0.0
@@ -530,7 +513,6 @@ types-s3transfer==0.6.1
 typing-extensions==4.10.0
     # via
     #   aioredis
-    #   asgiref
     #   azure-core
     #   azure-keyvault-secrets
     #   azure-servicebus
@@ -556,6 +538,7 @@ typing-extensions==4.10.0
     #   sqlalchemy
     #   starlette
     #   typing-inspect
+    #   uvicorn
 typing-inspect==0.9.0
     # via dataclasses-json
 tzdata==2023.3
@@ -572,7 +555,7 @@ urllib3==1.26.16
     #   kubernetes
     #   kubernetes-asyncio
     #   requests
-uvicorn==0.17.6
+uvicorn==0.30.0
     # via -r model-engine/requirements.in
 uvloop==0.17.0
     # via -r model-engine/requirements.in