Concurrency

asyncio, threading, multiprocessing — and when each is actually the right tool given the GIL.

Why we need this / what value this brings

Localz's FastAPI backend mixes sync and async code paths; picking the wrong concurrency primitive silently kills throughput.

When to use this

I/O-bound work (asyncio), CPU-bound work (multiprocessing), or blocking legacy code inside an async app (threads via run_in_executor).

How to use or implement this

See the general concurrency theory first, then this folder for Python-specific application. The Concurrency and Scalability Checklist for async Python lives here too.

Research questions

• (add as research surfaces open questions)

---

Empty folder — drop notes, links, and findings here as you research.