FastAPI CI/CD Pipelines — Ship Your Async Python API at Full Speed

FastAPI's async nature introduces testing complexity. Async test functions need pytest-asyncio, async database sessions require careful fixture management, and httpx.AsyncClient replaces the synchronous test client for realistic testing. Getting these configured correctly in CI is non-trivial.

Pydantic v2's migration changed how models validate data, and your CI pipeline should catch schema mismatches before they hit production. Type checking with mypy, combined with Pydantic's own validation, provides a strong safety net — but only if your pipeline actually runs these checks on every commit.

FastAPI applications typically use SQLAlchemy with Alembic for database migrations. The async SQLAlchemy engine adds complexity to migration scripts, and Alembic's autogeneration can miss certain changes (like column type modifications or enum updates). Your pipeline needs to verify that migrations are complete and reversible.

We configure pytest with pytest-asyncio in auto mode so every async test function is automatically detected. Test fixtures create async database sessions using async_sessionmaker with transaction rollback for isolation. The test client uses httpx.AsyncClient with FastAPI's app for realistic request/response testing including middleware.

The pipeline runs mypy with the pydantic plugin for accurate type checking of models and dependency injection. Ruff handles linting and formatting in a single fast pass. We also run alembic check to verify that no model changes are missing migration scripts — this catches the common mistake of modifying a SQLAlchemy model without generating a migration.

Docker builds use python:3.12-slim with a multi-stage build. The production entrypoint runs uvicorn app.main:app --host 0.0.0.0 --workers 4 with the number of workers tuned to the container's CPU allocation. We configure Gunicorn as the process manager with Uvicorn workers for production-grade process management, including graceful shutdown handling and worker recycling.

A FastAPI-specific CI/CD pipeline:

• Async test automation — pytest-asyncio with httpx.AsyncClient and database fixtures
• Type safety — mypy with Pydantic plugin on every commit
• Migration validation — Alembic check ensuring no missing migrations
• Fast linting — Ruff for linting and formatting in under 2 seconds
• Production Docker image — Gunicorn + Uvicorn workers, optimized for async
• OpenAPI validation — auto-generated docs verified as part of the pipeline
• Dependency management — Poetry or uv with lockfile-based installs