About

I'm a software engineering manager who still builds.

I lead an engineering team at a logistics-tech company, where I split my time between hands-on architecture and people leadership. Lately, most of that architecture work has been AI: I built the Claude/LLM extraction layer that moved one of our core products from prototype to a production pipeline, and I designed an internal Claude Code platform — skills, agents, and reference docs — that codifies how our team plans, builds, and reviews work with AI assistance. It's since been picked up by teams beyond my own.

The part I keep coming back to is efficiency. The same discipline that makes an AI system cheaper to run — tighter context, the right model for the job, less wasted computation — also makes it lighter on energy and water. That intersection of AI efficiency and sustainability is underexplored, and it's most of what I write about here: what it actually takes to get an LLM from demo to dependable production, and what it costs — in dollars and in resources — to keep it running well.

This site is where I think out loud about that work: real problems, how I approached them, what held up and what didn't.

If you want to get in touch, see the Contact page.