I build data systems and cloud architecture at AWS · Seattle
Six-plus years turning ambiguous business problems into production data platforms, most recently a performance analytics ecosystem used across multiple AWS organizations. On my own time, I build the products I find interesting; the rest of it goes to tennis, books, and people I like. AWS Solutions Architect certified.
Production systems I've designed and shipped, and what made each one hard.
A four-application performance analytics ecosystem on one shared data foundation: 70+ metrics per person, serving individual contributors, managers, and operations across multiple AWS business units. Shipped in a seven-week sprint; now the designated data source for official review cycles. I was the architecture partner on the data layer.
A production event platform where every tenant's data lives in the same PostgreSQL database, isolated by row-level security and JWTs validated at the edge. The design got its trial by fire when an implicit type mismatch broke RLS policies in production. The full story is in the case study.
The assistant on this site, built as a study in cost-bounded serverless design: per-user and global rate limits in DynamoDB, persistent conversation memory, zero fixed infrastructure cost. It took three architecture rebuilds to get there. Ask it how it works; it knows its own design.
Six specialized XGBoost models pick my Fantasy Premier League squad and captain every gameweek, retrained automatically on a pipeline that needs no hand-holding. It also runs a public scoreboard of its own accuracy, gameweek by gameweek, no matter what it shows.
Technical essays and personal pieces.
Studying, building, reading — here's what's on right now.