I started at Imperial College London studying Mining Engineering, which turned out to be excellent preparation for a career in software — both involve a lot of time sitting in a dimly lit place, hitting things until they work, and hoping the physical or metaphorical ceiling holds. I didn't finish the degree. Instead I drifted into teaching people how to do computer things and writing code in the late nineties, first in London, then Rotterdam, then Brisbane, then eventually settling in Blaine, Washington, a small town on the Canadian border whose main distinction is that it's slightly easier to leave than most places. I've been building software professionally for nearly thirty years, which mostly just means I've been wrong about things for longer than most people have been alive.

The work has ranged from consulting for NASA JPL, Lockheed Martin, and Walmart through my own shop, to co-founding an open source IDE plugin that somehow got adopted internationally, to spending a decade and a half at T-Mobile doing architecture work of varying ambition and questionable quality. Two patents emerged from that period, which sounds more impressive than it is — patents are mostly just proof you wrote something down before someone else did. More recently I've been deep in agentic AI — MCP servers, multi-agent pipelines, the works. I find it genuinely interesting, which is either a good sign or evidence that I've been doing this too long and have lost perspective. Possibly both.

This blog is mostly a place to write about building things. The newsletter is generated by software I built to avoid having to synthesize my reading manually, which is either clever automation or an elaborate way to procrastinate. I haven't decided yet.