I document my engineering journey—the problems I solve, the systems I build, and the lessons I learn. Every day brings new challenges in frontend development, system design, and shipping code at scale.
⋆⊱༻𖥸༺⊰⋆
Debugging the boundary between an Elixir service and a Bedrock-hosted Claude, where a wrong hostname, a wrong version string, and a Markdown code fence all look identical from the outside.
Compressing a large PR into a token budget is easy; the hard part is choosing which files to discard and admitting to the model that you did.
A missing request timeout, a tracked scratch directory, and a swallowed git error left an automated blog agent hung for 11 days and running stale code for a week before that.
A deep dive into building Zacian, an automated code review bot for GitHub using Claude AI. Learn about the architecture, multi-stage verification pipeline, token accounting, and how to handle concurrent AI workloads in Elixir.
A quality scorer that ignores its own standards, a rewrite step that throws its output away, and a grep that kills the run before either matters — found by pointing the agent at itself.
A complete guide to building self-directed agents that autonomously discover, analyze, and publish technical blogs from daily engineering work
Page 5/6