Gemma 4 on an Amazon SageMaker Endpoint: AWS CLI, NVIDIA L4, and an MCP Server

This article provides a step by step deployment guide for Gemma 4 E2B to an Amazon SageMaker hosted GPU enabled system. A suite of Python MCP tools is built to simplify management of the vLLM hosted deployment with Claude Code. https://github.com/xbill9/sagemaker-gemma What is this project trying to Do? This project serves Gemma 4 E2B from a SageMaker real-time endpoint on one NVIDIA L4 GPU, using the vLLM container AWS publishes for SageMaker. Every AWS call is a plain aws CLI command, so each step can be run by hand or by the MCP server. A SageMaker real-time endpoint is a managed HTTPS inference server. SageMaker places the container on a GPU instance, health-checks it, routes requests to it and writes its logs to CloudWatch. There is no instance to patch, no security group to open and no load balancer to build. Where do I start? The strategy for starting MCP development for model management is a incremental step by step approach. First, the basic development environment is setup with the required system variables and a working Claude Code configuration. Then, the Python MCP server is brought up over stdio and validated with Claude Code in the local environment. The deployment follows as eight steps, each shown as the raw aws command and the MCP tool that runs it. At This Point You Should Have… - An AWS account and the AWS CLI v2, signed in with aws login - A SageMaker endpoint quota of at least 1 for a single-L4 instance type ( ml.g6.xlarge ,ml.g6.2xlarge orml.g6.4xlarge ) - Python 3.11 or newer with mcp 2.x - Claude Code or Gemini CLI installed and working - jq for reading JSON replies Setup the Basic Environment Clone the repository and install the one requirement into the system Python: git clone https://github.com/xbill9/sagemaker-gemma cd sagemaker-gemma python3 -m pip install -r requirements.txt cp .env.example .env .env is gitignored. It holds the settings every tool reads: cat .env # Copy to .env (gitignored) and edit. sm.py reads it, so the MCP server, CLI and Makefile all see it. AWS_REGION=us-east-2 MODEL_ID=google/gemma-4-E2B-it INSTANCE_TYPE=ml.g6.xlarge ENDPOINT_NAME=gemma-4-e2b ROLE_NAME=sagemaker-gemma-execution-role MAX_MODEL_LEN=8192 # Leave empty to use the newest SageMaker vLLM image in the region. IMAGE_URI= # Fallback instance types, highest priority first (same GPU keeps runs comparable). INSTANCE_POOLS=ml.g6.xlarge,ml.g6.2xlarge,ml.g6.4xlarge Gemma 4 is Apache-2.0 and ungated on Hugging Face, so no Hugging Face token is needed. Model Management Tool with MCP Stdio Transport The simplest MCP transport is stdio: the client launches the server as a local process and talks to it over stdin and stdout. In this project Claude Code is the MCP client. The server is one file, server.py , on the MCP Python SDK 2.x: mcp = MCPServer(RIG_NAME) READ_ONLY = ToolAnnotations(readOnlyHint=True, idempotentHint=True) WRITE = ToolAnnotations(destructiveHint=False) DESTRUCTIVE = ToolAnnotations(destructiveHint=True) Every tool carries one of the three annotations, so a client can tell a status check from a deploy from a delete. The tools do no AWS work themselves. They call sm.py , which runs each request as an aws CLI subprocess: def aws(*args: str, region: str | None = REGION, parse: bool = True) -> Any: """Run aws --output json and return the parsed result.""" cmd = ["aws", args, "--output", "json"] The CLI renews an aws login session on its own, so a server that stays up for hours keeps working credentials. sm.py also drops any AWS_SESSION_TOKEN the server inherits from its parent process, because a static token expires inside a long-running server and outranks the login session. Running the Python Code The project can be linted: make lint All checks passed! 16 files already formatted and tested: make test ---------------------------------------------------------------------- Ran 16 tests in 0.010s OK The tests replace the aws subprocess with a fake, so they run offline with no credentials. One of them compares the registered tool set and annotations against a fixed list: a tool that failed to register, or a delete that lost its destructive flag, fails the suite. ✅ Test the Protocol by Hand A client speaks JSON-RPC over stdio. Hold stdin open with sleep , or the server sees end-of-input and exits before it answers: { printf '%s\n' \ '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"probe","version":"0"}}}' \ '{"jsonrpc":"2.0","method":"notifications/initialized"}' \ '{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}'; sleep 3; } \ | python3 server.py 2>/dev/null Summarised: 1 sagemaker-gemma 2 ['check_quotas', 'delete_endpoint', 'deploy_endpoint', 'find_vllm_image', 'get_deployment_config', 'get_endpoint_logs', 'get_endpoint_status', 'get_help', 'list_endpoints', 'query_model', 'verify_model_health'] 🟢 The server answers the handshake and lists 11 tools. Claude Code .mcp.json Claude Code reads .mcp.json in the project directory and launches the server with the system python3 : { "mcpServers": { "sagemaker-gemma": { "command": "python3", "args": ["/home/xbill/sagemaker-gemma/server.py"] } } } Gemini CLI reads the same entry from .gemini/settings.json . Validation with Claude Code claude mcp get sagemaker-gemma sagemaker-gemma: Scope: Project config (shared via .mcp.json) Status: ✔ Connected From inside Claude Code, get_help returns the resolved settings and the order of work: ### sagemaker-gemma Serving google/gemma-4-E2B-it with vLLM on a SageMaker real-time endpoint, managed entirely through the aws CLI. | Setting | Value | | --- | --- | | Region | us-east-2 | | Endpoint | gemma-4-e2b | | Instance types (priority order) | ml.g6.xlarge, ml.g6.2xlarge, ml.g6.4xlarge | | Max model length | 8192 | | Image | newest SageMaker vLLM image (find_vllm_image) | | Execution role | sagemaker-gemma-execution-role | Order of work: check_quotas → deploy_endpoint → get_endpoint_status until InService (about 10 minutes once an instance is placed) → verify_model_health → query_model → delete_endpoint. The steps below follow that order. Each one shows the raw CLI command; deploy_endpoint runs Steps 2 to 6 in one call. export AWS_REGION=us-east-2 AWS_PAGER= export NAME=gemma-4-e2b export MODEL_ID=google/gemma-4-E2B-it Step 1 - Check the Endpoint Quota SageMaker endpoint quotas are per instance type, per region, and count instances. aws service-quotas list-service-quotas --service-code sagemaker \ --query "Quotas[?QuotaName=='ml.g6.xlarge for endpoint usage'].Value" [ 1.0 ] The check_quotas tool reads the same quota for each fallback type across the US regions: | Instance | us-east-1 | us-east-2 | us-west-1 | us-west-2 | | --- | ---: | ---: | ---: | ---: | | ml.g6.xlarge | 1 | 1 | - | 1 | | ml.g6.2xlarge | 1 | 1 | - | 1 | | ml.g6.4xlarge | 1 | 1 | - | 1 | Regions with quota for at least one of these types: us-east-1, us-east-2, us-west-2. A - means the type is not offered in that region. A 0 means a quota increase request first. 🔎 Tip: One Quota Call per Region list-service-quotas pages through every SageMaker quota in the region, and the Service Quotas API is rate limited per account. Twelve calls at once, one per type per region, came back as: An error occurred (TooManyRequestsException) when calling the ListServiceQuotas operation (reached max retries: 2) check_quotas makes one call per region, one region at a time, and filters the types from that one result. sm.py also sets AWS_RETRY_MODE=adaptive so the CLI backs off and retries. Step 2 - Find the vLLM Container The AWS vLLM repository holds pinned release tags such as 0.30.0-gpu-py312-cu130-ubuntu24.04-sagemaker-v1.1 , floating aliases such as 0.30-gpu-py312 , and -soci index tags. Older release lines receive patch rebuilds, so the most recent push can carry an older vLLM. Sort the pinned tags by version: TAG=$(aws ecr describe-images --registry-id 763104351884 --repository-name vllm \ --query "imageDetails[].imageTags[]" --output text | tr '\t' '\n' \ | grep -E -- '^[0-9]+.[0-9]+.[0-9]+-.-sagemaker-v[0-9]+.[0-9]+$' | sort -V | tail -1) export IMAGE=763104351884.dkr.ecr.$AWS_REGION.amazonaws.com/vllm:$TAG echo $IMAGE 763104351884.dkr.ecr.us-east-2.amazonaws.com/vllm:0.30.0-gpu-py312-cu130-ubuntu24.04-sagemaker-v1.1 The find_vllm_image tool applies the same rule. The image runs vLLM 0.30.0. Step 3 - Create the Execution Role SageMaker assumes this role to pull the image and write logs. It is created once per account. aws iam create-role --role-name sagemaker-gemma-execution-role \ --assume-role-policy-document '{"Version":"2012-10-17","Statement":[{"Effect":"Allow","Principal":{"Service":"sagemaker.amazonaws.com"},"Action":"sts:AssumeRole"}]}' aws iam attach-role-policy --role-name sagemaker-gemma-execution-role \ --policy-arn arn:aws:iam::aws:policy/AmazonSageMakerFullAccess export ROLE=$(aws iam get-role --role-name sagemaker-gemma-execution-role --query Role.Arn --output text) On a second run create-role reports EntityAlreadyExists ; get-role still sets ROLE . deploy_endpoint creates the role only when it is missing. Step 4 - Create the Model A SageMaker model pairs an image with its settings. The container turns each SM_VLLM_ variable into the matching vLLM flag: SM_VLLM_MAX_MODEL_LEN becomes --max-model-len . aws sagemaker create-model --model-name $NAME --execution-role-arn $ROLE \ --primary-container "{"Image":"$IMAGE","Environment":{ "SM_VLLM_MODEL":"$MODEL_ID", "SM_VLLM_MAX_MODEL_LEN":"8192", "SM_VLLM_GPU_MEMORY_UTILIZATION":"0.9"}}" { "ModelArn": "arn:aws:sagemaker:us-east-2: :model/gemma-4-e2b" } Step 5 - Create the Endpoint Config With Fallback Instance Types The endpoint config says where the model runs. InstancePools lists up to five instance types in priority order, and SageMaker places the first one with a free instance. All three types below carry one L4, so the model sees the same GPU whichever is placed