Skip to main content
You can deploy BAML applications to AWS Lambda using SST to define the Lambda configuration.

​
Node.js + SST Example

The example below builds the BAML x86_64 Rust binaries into a Lambda layer and uses the layer in the Lambda function.

Node + AWS Lambda + SST Example

View the complete example project on GitHub

​
Key Components

The example demonstrates how to:
  1. Build BAML binaries - Package the BAML runtime as a Lambda layer
  2. Configure Lambda - Set up the Lambda function with the appropriate runtime
  3. Deploy with SST - Use SST to manage the deployment process

​
Lambda Configuration

The BAML binaries are built for x86_64 architecture and packaged as a Lambda layer. Your Lambda function references this layer to access the BAML runtime.
Example SST Configuration
import { Api, StackContext } from "sst/constructs";

export function MyStack({ stack }: StackContext) {
  const api = new Api(stack, "api", {
    routes: {
      "POST /generate": "functions/generate.handler",
    },
  });

  stack.addOutputs({
    ApiEndpoint: api.url,
  });
}

​
Runtime Requirements

The BAML binaries only support the Node.js 20.x runtime (or a runtime using Amazon Linux 2023).If you need a different runtime version, please let us know.

​
Python Support

We’re working on a Python example for AWS Lambda deployment. If you need to deploy a Python BAML project on AWS, please reach out to us on Discord or GitHub.

​
Environment Variables

When deploying to AWS Lambda, ensure you configure the necessary environment variables:
  • LLM API Keys - Set your OpenAI, Anthropic, or other provider API keys
  • BOUNDARY_API_KEY - Optional, for Boundary Studio integration
  • BAML_CACHE_DIR - Optional, to customize the cache directory location
You can configure these in your SST configuration or through the AWS Lambda console.
Setting Environment Variables in SST
import { Function, StackContext } from "sst/constructs";

export function MyStack({ stack }: StackContext) {
  const fn = new Function(stack, "MyFunction", {
    handler: "functions/generate.handler",
    environment: {
      OPENAI_API_KEY: process.env.OPENAI_API_KEY!,
      BOUNDARY_API_KEY: process.env.BOUNDARY_API_KEY!,
    },
  });
}

​
Best Practices

  1. Use Lambda Layers - Package BAML binaries as a layer to reduce deployment size
  2. Cache Dependencies - Leverage BAML_CACHE_DIR for faster cold starts
  3. Monitor Performance - Use Boundary Studio to monitor LLM calls and performance
  4. Secure Secrets - Store API keys in AWS Secrets Manager or Parameter Store

​
Next Steps

Build docs developers (and LLMs) love

Get started for freeTalk to us