Python API Stack
Verdict
Preferred — FastAPI + Pydantic + Uvicorn is the standard Python API stack; adopt all three together, not as separate picks.
Use when
- A Python backend has been justified under the criteria in Python Toolchain.
- Building a Python API that needs data validation and structured models.
- Serving a FastAPI application whose deployment platform provides no managed ASGI server.
Avoid when
- A TypeScript-only architecture is sufficient — do not introduce a Python backend automatically (see Python Toolchain).
- The deployment platform provides an appropriate managed alternative to Uvicorn.
Strengths
- FastAPI: high-performance ASGI framework with automatic OpenAPI documentation.
- Pydantic: typed data validation and structured models that integrate with FastAPI request and response handling.
- Uvicorn: lightweight, standard ASGI server for serving FastAPI applications.
Tradeoffs
- Adds a separate Python service to manage and deploy.
- Pydantic models must be kept in sync with the data they validate.
Version policy
- Prefer the latest stable versions of FastAPI, Pydantic, and Uvicorn for new projects.
- Do not hold the stack on obsolete versions without a concrete compatibility reason.
Usage rules
- Use FastAPI as the Python API framework.
- Use Pydantic for Python data validation and structured models where appropriate.
- Use Uvicorn as the standard ASGI server for FastAPI applications unless the deployment platform provides an appropriate managed alternative.
Works with
Internal interplay
- FastAPI + Pydantic — Pydantic models are FastAPI's validation layer; define request and response schemas as Pydantic models rather than hand-rolled validation.
- FastAPI + Uvicorn — Uvicorn is the standard ASGI server for FastAPI applications unless the platform provides a managed alternative.
Related guidelines
- Python Toolchain — pairs-with: manage environments, dependencies, and linting for this stack with uv and ruff.