Proof of Building
Things I've built, shipped, or explored.
The clearest current proof starts with AI Evals Lab and AI Research Residency. One is a compact interactive product about eval posture and failure buckets. The other is the longer research track around transformers, benchmarks, agent systems, and public learning proof.
The rest of the newer cluster is a set of small interactive products around AI product taste, recruiting fit, workflow design, interface shape, and eval posture. I am trying to keep this set tighter now so the strongest ones can become more real products instead of a pile of adjacent ideas.
For the shortest route in, start with Interactive AI Products. For a direct explanation of how Claude Code and Codex show up in the process, read what I've actually built with Claude Code and Codex. For the research side, including evals, benchmarks, and agent systems, read this note.
AI Evals Lab
Interactive eval-design simulator for choosing benchmark posture, failure buckets, review depth, and the right release artifact for an AI workflow.
AI Research Residency
Public AI course and research residency covering transformers, evals, benchmarks, agent systems, and proof-of-mastery artifacts across code, notes, and experiments.
AI Product Reactor
Interactive simulator for shaping AI products people want to touch, share, and return to, with guidance on the first experience, retention loop, and product surface.
AI Company Fit Radar
Interactive simulator for deciding where an AI builder or operator fits, what proof artifact to send, and how to frame outreach to AI labs and AI-native teams.
AI Workflow Lab
Interactive workflow simulator for tuning model freedom, eval rigor, review depth, and the first proof artifact worth shipping.
AI Interface Lab
Interactive AI product simulator for choosing the right interface shape, trust boundary, wow moment, and first release for an AI-native product.
Compounder
Privacy-first habit tracker. No signup, no cloud, no data harvesting. Just you and your streaks. 1,000+ active users.
Bookmark
AI-curated reading digest. Three hand-picked articles every Monday morning, summarized and delivered to your inbox.
The List
50 tech companies worth paying attention to: AI labs, dev tools, infrastructure. Opinionated commentary, updated monthly.
MCP Studio
MCP-powered Airtable client and internal tool demo for AI workflow orchestration, operator tooling, and hands-on product exploration.
AI Streaks
AI product and workflow tool for turning AI-generated plans into guided 7-day execution and operator follow-through.
MBA Copilot
RAG-based AI study copilot built with Next.js, FastAPI, Pinecone, and OpenAI for coursework, retrieval, and AI-assisted learning workflows.
Voice AI Assistant
Conversational voice AI browser game prototype where an OpenAI-powered retention agent tries to stop you from cancelling a subscription.
Agentic Payments
Market map and research product covering agentic commerce, payments, identity, wallets, and protocol infrastructure.
Forecast Desk
Real-time weather dashboard built on National Weather Service data. Hourly and 7-day forecasts for New York.