🧠 One of the most underused yet insanely powerful Python tools for performance optimization — the Least Recently Used (LRU) cache. If your code recomputes expensive function calls, this decorator can change everything.

How I cut my CI/CD pipeline from 12 minutes to 4 minutes. Changes: - Parallelised long running test jobs (i.e. end-to-end tests with selenium). - Properly cached external dependencies installation (i.e. npm install / uv sync) and Docker layers. GitLab runner costs dropped 60%.

I’ve just tried using Pyrefly and Ty in a python project with an existing codebase. The Rust based type checkers run circles around mypy and pyright, and Purefly does seem stricter than Ty. A small example: