Writing

Things I keep having to explain.

Mostly arguments I have had more than once — about what a system should give up under load, why a cache needs hermetic inputs, and what makes an AI pipeline safe to keep.

18 May 2026

Degrade in latency, not in correctness

When a system saturates it has to give something up. Almost every default gives up the wrong thing — and users only find out much later.

Distributed systemsReliabilityBack-pressure

6 min

27 March 2026

Your AI system needs an evaluation suite before it needs a better prompt

Prompt engineering without measurement is guessing with extra steps. The suite is what turns a demo into something you can safely change.

AI automationEvaluationProduction ML

7 min

22 January 2026

Hermetic builds are not an optimisation

Remote caching on a non-reproducible build does not make it faster. It makes it confidently wrong, and I learned that the expensive way.

Build systemsCI/CDReproducibility

5 min

9 November 2025

The pull request is the right unit of infrastructure

If getting an environment means asking a person, your platform has a queue in it. The fix is to make the pull request provision its own world.

Platform engineeringKubernetesDeveloper experience

6 min