Distributed systems
Consensus, sharding, backpressure, and the unglamorous work of keeping services up at 3 a.m. GMT.
Twelve years building infrastructure that served a few billion people across Google, Meta and Amazon. Now I write from London, advise early-stage founders, and help engineers get through the hard middle part of their career.
Three things, mostly. Everything I publish is something I wish I'd read five years earlier.
Consensus, sharding, backpressure, and the unglamorous work of keeping services up at 3 a.m. GMT.
Interviews, promotions, feedback, and the organisational politics nobody teaches you in a bootcamp.
Using LLMs sensibly, shipping evals, and separating real workflow gains from LinkedIn theatre.
A rotating selection from the archive. The full list lives on the blog page.
Five uncomfortable things I learned after leaving big tech — and what I'd tell my 25-year-old self.
A repeatable structure I use for any system design question, from URL shorteners to global payments.
The architecture decisions behind a system I helped build that served 1B+ DAUs, minus the marketing gloss.
The messy, honest version of why I walked away from a seven-figure compensation package.
Why most teams don't need them, and how I've watched three companies walk themselves into a distributed monolith.