Product leader based in San Francisco. I started as a semiconductor engineer designing motor control and power electronic systems. Everything since has been building things where the hard part isn't the technology. It's making it disappear, and delighting the users.
From LiDAR traffic intelligent safety infrastructure, to ML for battery analytics, Mobile EV charging, Grid orchestration for distributed energy, space awareness and network management for contained sovereign environments. Now AI systems for strategic work.
The throughline: absorb the complexity so the person using it doesn't have to.
Atelier Atelier is the first system to learn and preserve an expert-owned standard as an authority-controlled target, then compile the minimum architecture needed to reproduce it reliably across repeated generation.
Give it work an expert considers excellent. It infers the standards behind that work, the expert ratifies what's actually true, and it compiles the result into a portable skill. The expert owns what good means. The tool owns the job of reproducing it, and can improve how it does that without ever changing what was approved.
Built because I kept hitting the same wall. I could give a model perfect context and still get variance. Knowing what information to feed a model turns out to be a different problem from controlling what standard it reproduces.
skills AI skills for Claude Code. Product marketing, strategy, GTM.
StratOS (private) A strategy operating system for early-stage teams.
Founders, PMs and PMMs make the same high-stakes calls with far too little evidence. Who is this for. What do we say. What do we charge. What do we build next. StratOS produces the strategic work behind those decisions, from positioning and pricing to PMF diagnosis and GTM motion, then tracks whether the bets you made are still true as the company moves. What used to take a quarter and an outside consultant takes an afternoon.
Models are getting cheaper at everything objective. What stays scarce is the part that was always hardest to write down: what to emphasize, what to ignore, what feels wrong despite looking reasonable, when to break the framework, which technically correct answer you'd never ship.
Most of my work is some version of turning that invisible layer into infrastructure.
I write about product, AI systems, energy, and what building hardware teaches you about building anything. TRULY ANYTHING!
Hardware taught me you can't hide behind iteration. You have to understand deeply to build simply.