infrastructure-patterns

To deploy Python and LLM applications, you can use Docker multi-stage builds with uv for efficient containerization, and Kubernetes deployments with Horizontal Pod Autoscalers to manage scalable LLM service delivery. This Skill provides the manifests and patterns for both platforms.

The best way to manage secrets for Python apps is by using Pydantic BaseSettings for application configuration, SOPS for encrypting secrets directly in Git, and integrating external secrets with Kubernetes to secure your CI/CD pipelines.

Yes, you can use uv for Docker multi-stage builds in Python applications to create efficient container images. This Skill provides specific Dockerfile patterns that leverage uv to optimize the build process and reduce image size.

You can set up GitHub Actions CI/CD pipelines for FastAPI and LLM services using comprehensive workflows that cover linting, testing, building, and deploying. This Skill includes caching strategies and deployment configurations for automated delivery.

You implement liveness and readiness probes for FastAPI by creating dedicated health check endpoints. This Skill provides FastAPI patterns to build robust health checks that Kubernetes uses to monitor application availability and route traffic correctly.

Yes, Docker Compose works effectively for local development and testing of LLM applications. This Skill provides Docker Compose configurations that allow you to replicate your production environment locally for consistent testing.