I work on the systems most people never think about until they break: queues, schedulers, and the databases underneath them. Nine years in, mostly at seed-to-Series-B companies where the infrastructure has to be built once and left alone.
Off the clock I maintain a couple of open-source CLIs and write short, practical notes on distributed systems — no hot takes, just what I wish I'd known before the incident.
A durable job scheduler for services that can't afford to lose a task: at-least-once delivery, exactly-once side effects via idempotency keys, and a dashboard that shows exactly why a job retried.
A leader-election and config-distribution layer built on Raft, extracted from three internal services that each rolled their own version badly.
A CLI that diffs a Postgres schema against a set of migration files and fails CI before the drift does.
An append-only event store with snapshotting, built for a payments team that needed a real audit trail, not just logs.
Own the job-scheduling and event-pipeline platform used by every product team. On call, by choice, for the systems I built.
Built the leader-election layer that became Quorum, and the services on top of it.
Payments infrastructure for a mid-market billing platform.
Devon is the engineer you want holding the pager. Waypoint replaced a system three of us were afraid to touch, and it just works.
contact --devon
