Deploy python applications using sst.

SST uses uv to manage your python runtime. If you do not have uv installed, you can install it here. Any sst workspace package can be built and deployed to aws lambda using sst. In this example we deploy an API handler to lambda from the functions directory. The handler depends on shared code from the shared directory using uv's workspaces feature. (Note: builds currently do not tree shake so lots of workspaces can make larger builds than necessary.)

Python functions can be deployed just like other SST functions, the only difference is that the functions themselves must be configured within a uv workspace, there is no drop-in-mode.

const python = new sst.aws.Function("MyPythonFunction", {
  handler: "functions/src/functions/api.handler",
  runtime: "python3.11",
  url: true
});

If you are using live lambdas for your python functions, it is recommended to specify your python version to match your Lambda runtime otherwise you may encounter issues with dependencies.

[project]
name = "aws-python"
version = "0.1.0"
description = "A SST app"
authors = [
    {name = "<your_name_here>", email = "<your_email_here>" },
]
requires-python = "==3.11.*"

Live lambda will locally run your python code by building the workspace and running the specified handler. You can have multiple handlers in the same workspace and have multiple workspaces in the same project.

.
├── workspace_a
│   ├── pyproject.toml
│   └── src
│       └── workspace_a
│           ├── __init__.py
│           ├── api_a.py
│           └── api_b.py
└── workspace_b
    ├── pyproject.toml
    └── src
        └── workspace_b
            ├── __init__.py
            └── index.py

Keep in mind that AWS Lambda zip archives have limits and python does use native extensions for certain packages, if you are using large dependencies such as numpy, pandas, and others found in the SciPy stack, you may want to use the container mode to deploy your python code.