I write about systems I can measure. That usually means a backend service, a database migration, a CLI tool, or a pipeline that failed in a way worth explaining. If a number appears in a post here, it came from an API, a test run, or a command whose output I can show.
This site is deliberately plain: static HTML generated from markdown files kept in a git repository. No database, no tracking scripts, no comments. The files are the source of truth, and a push rebuilds the published site.
What gets published here
- Notes on backend work in Go, TypeScript and Python, mostly on PostgreSQL.
- Post-mortems of things that broke, with the fix and the reasoning.
- Write-ups of the open-source tooling I maintain, including agentdiet, a CLI that measures the context an AI coding agent loads on every request and fails a build when it grows past a budget.
- Migration and delivery work: what moved, what it cost, and how it was verified.
Where else this appears
Longer pieces are cross-posted to dev.to. The canonical URL always points back here, so search engines and readers land on one copy.
That is the whole plan. Publish when there is something measured to say, and skip the week when there is not.