I build privacy-first software. After 13+ years across AI infrastructure, distributed systems, and machine learning, I've come to believe the most interesting work right now is making AI run entirely on the user's device — fast, private, and cheap to operate, with no servers to babysit.
I split my time between independent product work — Apple-platform apps focused on running language models locally — and a small amount of consulting for teams shipping on-device ML features.
The thread connecting everything is a single bet: that as small models get good enough, the right place to run them is the device they're already on, not a datacenter the user can't see. It's faster, more private, and — once you've felt it — surprisingly hard to give up.
13+ years of software engineering, primarily focused on AI and distributed systems. Experience spans:
Currently applying that background to consumer apps where on-device inference replaces network calls, and the same engineering rigor used in serious AI architecture goes into tools people use daily.
Daily drivers: Swift, Rust, llama.cpp, MLX, Metal.
Comfortable across Python, C++, TypeScript, and Go when the situation calls for them. Long history with C# and Java from earlier in my career.
I take on a small number of consulting engagements each year — typically helping teams architect on-device ML features, optimize Apple Silicon inference, or design Swift ↔ Rust bridges. Engagements are usually 4–8 weeks.
If that sounds useful, send a note via the form with a sentence or two about what you're building. I read every message and reply within a couple of days.