Essays and notes on software engineering — distributed systems, clean architecture, developer experience, and lessons from shipping real production code.
An AI agent that opens one microservice only sees that microservice — and fills the gaps with fiction. Here's the pattern I use to give agents (and humans) a shared, cheap-to-load, validated map of the whole ecosystem.
In 2017, before LLMs made enterprise AI mainstream, I built a multi-channel chatbot that automated L1 support at scale. Here's the architecture, the results, and what it looks like rebuilt with today's tools.
Logs alone won't save you at 3am. Here's the observability setup I use to make distributed systems debuggable by design — structured logs, distributed traces, and alerts that actually mean something.
How to keep payment and order flows consistent across microservices at scale — with real failure modes, idempotency requirements, and SQS-backed orchestration that degrades gracefully instead of deadlocking.
From $20,000 to $1,800/month — no infrastructure change, no migration to another service. Just a better data model.
One question surfaces more production risk than any checklist: what happens when this fails? Here's how to ask it, answer it with concrete patterns, and prove your answers hold under real load.
How to cut cluster costs by 40–70% by replacing guessed resource configs with metric-driven right-sizing — without sacrificing reliability.
From 2 minutes to 2 seconds for 924 messages — two targeted changes I made to our messaging layer, no infrastructure swap required.
How I migrated a decade-old Subversion repository to Git without losing a single commit, unlocking a pipeline rewrite that took customer releases from more than a day to ~1 hour — 87% faster.