Senior software engineer — applied AI, product, and customer-facing. Nine years building products end to end: iOS, backend, and now production AI systems. These days I split my time between SDK engineering, helping enterprise customers' engineering teams ship complex integrations, and wiring AI agents into real production loops. I care more about shipped outcomes than clever code, and I do my best work embedded with a team getting something to actually work.
Mostly: Swift · TypeScript · Rust · agent frameworks & harnesses (Claude Code, Codex, Pi) · AWS / serverless · eval loops & multi-model setups since 2023.
🔭 I’m currently working on descartes (today an ops-triage agent, tomorrow your machine's guardian angel and your main ops interface. But I'm starting with the skateboard version, the rocketship will come later), Boucle is building itself, and I also work on third thoughts (because there's so much to learn about how to interact with LLMs and agents), the foundry plugin (adversarial development with information boundaries), and a few closed-source iOS apps.
🌱 I’m currently learning how to run 10+ agents simultaneously on just as many projects without going mad. Next step looks like it will be running 10+ agents each orchestrating 10+ agents per project on 10+ projects. GasCity, here I come.
📝 I write at The (not so) Weekly Fitz
👯 I’m looking to collaborate on interesting and unconventional LLM and agent applications.
🤔 I’m looking for help with descartes and third thoughts. Testers are welcome, and so is anyone with hands-on experience in privacy, privacy-conscious federated knowledge aggregation, or systems.