I lead AI and machine learning teams, and have been building production ML systems since 2014. Lately that means multi-agent LLM systems, which is a grand way of saying I spend most of my time on the unglamorous questions: is it safe, is it actually right, and will anyone be able to maintain it in six months. Before that I worked on story recommendations, loyalty offers, financial risk models, news auto-tagging, and social media analytics for DARPA, which is a career path I would not recommend planning but can recommend having. I like walking into a domain I know nothing about and figuring out how to ship something in it.

I got here sideways. Born in New Brunswick, raised in Bermuda, then fifteen-odd years in Japan, where I arrived to teach English and somehow left as a university professor researching computational linguistics. I have four degrees and started university in engineering before defecting to philosophy and literature, which I maintain was excellent preparation for arguing about system design. Years ago, before the CBC offered podcasts, I wrote a script that woke up whenever my favourite radio show aired, recorded the stream, tagged it, and dropped it into my music library. I have never really stopped doing that sort of thing: more recent evenings have gone into semantic search over a long-running film review podcast and a concept knowledge graph built out of Wikipedia. I am drawn to complexity, change, and a bit of chaos, and I would rather ship something useful now than something perfect later.