About

I ship production agents, generative creative pipelines (video, banner, HTML), and the infrastructure around them. At Liftoff I lead that work across sales tooling, creative production, and ad ops—taking new model capabilities from prototype to something teams can run every day.

Ten years shipping software: five years at Google on recommendations and surfaces at YouTube scale, two and a half years at WOMBO productionizing diffusion and video models for tens of millions of users, and since 2024 at Liftoff building GenAI engineering from the ground up.

Half my career predates calling a frontier model from an API. Latency, permissions, cost, observability, and review workflows were already part of the job. That is the lens I bring when the hard part is the gap between "this works in a notebook" and "this works on a Tuesday at 3am."

AI engineering is integration work with production consequences.

Model capability gets most of the airtime; integration gets too little. Most companies learn that through incidents. The teams that learn it earlier move faster.

Right now I'm thinking about how creative production changes when generation is cheap, how agentic systems behave for non-engineer users, and what the natural unit of work is for AI-driven workflows. I write about most of that on the blog.

I live in Austin, Texas. I play beach volleyball, tennis, and hockey, work out of local coffee shops most days, and read more than is probably healthy. Recent reading: Nexus by Yuval Noah Harari, The Avatamsaka Sutra (Cleary translation), and Why Nations Fail.