Five years on distributed, event-driven systems — Twilio today, Cimpress before that. Serverless on AWS, hard reliability problems, and the unglamorous work of making incidents stop repeating.
Most systems don't fail loudly. A queue backs up, a retry loops, and half the flows quietly stop working. I find that class of bug, fix the root cause, then add the alert that would have caught it — so nobody has to be clever about it again.
Four things I owned end to end — the problem, the constraint that made it hard, and the number that moved.
Message handling, retries, idempotency, and reprocessing workflows that recover data instead of dropping it.
Lambda, SQS, DynamoDB, Fargate, Terraform. Migrations that remove servers and the manual steps around them.
Logs, metrics and alerts that answer a question. Test coverage where regressions actually happen.
React and TypeScript when a system needs a face — including this site.
Reliability of distributed event-driven services: root-caused a production issue affecting half the transactional flows, built fault-tolerant reprocessing for failed async events, and raised coverage where regressions kept landing.
Built the inventory management system, the Merchant Center feed pipeline, and an integrated AWS chatbot that saved $500k+ in care costs. Mentored 3–4 engineers through reviews and design discussion.
Serverless migration of phone orders, PCI-compliant payment portals, and a competitor price-tracking system feeding Snowflake for predictive pricing (+15% revenue).
PL/SQL logic to categorize data into risk categories, an alert system that removed manual monitoring, and automated data loads from Smart Sheets into the database — the first taste of shipping into production.
Also: Golden Pen Award (National Pen, 2024) and 1st place with my team at CRIO Summer of Doing, 2019.
Open to senior backend and platform roles, and to short freelance work on reliability or serverless migrations. Fastest reply is email.