Here are things that I have worked on, including personal projects, writing, and public-facing work. If anything interests you, feel free to reach out.
2026
- Alpha Generation P1 (Self-Editing Heartbeats with Seeds): I think there are nontrivial ways to use current generation models to uncover alpha. Here is one method I have been using for the past few months. See the tweet here.
- Thrive Holdings Evals: Long-term holding companies are best positioned to understand the diffusion of AI and get at the question "Where is the ROI?" This set of benchmarks is a constantly changing artifact of our internal experience deploying models to create economic value.
- Mispricings in Young Talent (draft): A draft I'm working on which argues that young tech talent is extremely overlevered in the Valley and the stigma around nontechnical folks needs to go away with the advent of high-capex, heavily operational businesses.
- Cued: A project my friend Theo Tarr and I worked on in spring 2026 which inverts context management for consumer agents for a proactive future. Instead of creating tools/connectors to sync with iMessage/LinkedIn/X, proactively sync all events into an on-device DB with projection tables per event category and let the agent use SQLite commands to filter and search through contacts and messages.
- Surfaces: A project I worked on in May 2026 to create liquid glass UI for small plugins like copy/paste, live Codex threads, live PRs, and other notable widgets. Think of this as a minimal Raycast with a plugin factory.
2025
- ThreeOneOne: A project I worked on with Bruce Tang and Caden Li at HackMIT in September 2025. We built a ranking algorithm / Twitter clone for public aggregation of city issues, with maps + routing.
- Raju, the Chess Robot: A chess-playing robot I built with Alex Boniske, Marco Schönert, Carra Wu, and Talha Khan in April 2025. We won one of the largest robot arm hackathons in the world.
- Japanese Toilets and Care vs Slop: Sanitation has historically been a necessity in America while a luxury good in the rest of the world. What does this tell us about America in the age of AI and care vs slop?
- Simulation Fever & Applied Progress Studies: Simulation as a service can be a powerful mechanism for driving civilizational change, and it is just catching on.