Field notes from software work, since 2009.

What I measure, break and fix: AI coding agents, Linux, networking and developer tooling. Each post starts from a real problem and ends with something you can reproduce.

100 posts. Nothing from 2019 to 2025; writing started again in November 2026.

Latest

  1. Can a 125B open model do my small dev work on one GPU?

    Qwen3.8-Flash-Next on rented 5070 Ti and 3090 boxes: quality on real tasks, tokens per second, and what that means for upgrading a home server.

  2. Claude Code: Handling the error "API Error: The response stopped arriving"

    Every once in a while, I see a yellow warning on my claude code session: Searching about this error, it is exactly as it says: the server has stopped sending data mid-response,…

From the archive

All 100 posts, by year