Staff engineer · distributed systems · developer tooling
Notes from the trenches on backend architecture, reliability, and the small decisions that compound. New write-ups most weeks.
Exactly-once is a lie — here's how to get effectively-once without losing your weekend.
Pool sizing, timeouts, and the failure modes nobody warns you about.
A lifecycle for flags so they don't outlive the feature by two years.
The three shapes that explain most latency problems.
Newsletter · RSS · Archive