Working on
patch-pilot— the scoring model for dependency prioritization needs rework after I found edge cases where patch-level bumps in core networking libs were getting scored too low- This site — finally making it public after three months of "I'll add one more thing first"
- A write-up on the incident from last month where a misconfigured readiness probe caused a cascading failure across three services; still collecting notes
Reading
- The Pragmatic Programmer for the third time — different things land at different points in a career
- The Cilium documentation, specifically the eBPF datapath section, because I still don't fully trust what I haven't read
Paying attention to
The gap between what observability tooling promises and what it actually delivers in a four-person ops team with real budget constraints. Not everything needs a distributed trace; sometimes you just need a log that says what actually happened.
Also: whether patch-pilot should grow a web UI or stay a pure CLI tool. Leaning toward CLI-only and a good JSON output format that other tools can consume.
Not working on
Rewrites. I have three half-finished rewrites of old projects that I keep convincing myself are "almost done." They're not. Freezing them until the active projects ship.
Last updated: May 2026