Lacking a unified view of revenue, margins, and cost in a way that produced real insight, I built a platform that joins historical data with the business logic I’ve developed, synthesizing it into a decision-ready surface. The result is visibility that didn’t exist before, layered with the context that turns analytics into decisions that move the P&L.

I wanted to understand how metal actually moves, so I taught myself sand casting from the basics. I cast a putter head out of aluminum and brass, and it took several attempts before one came out clean. One sits on my desk, and the other lives in my bag.

I wanted an ambient wall display that I could talk to, and nothing on the market did what I had in mind. I taught myself Electron and built it. Voice-controlled, with a named zone layout pulling Spotify, stocks, news, weather, and sports, running on the wall in my space right now.

I wanted a pegboard that could change when my setup did, and none of the off-the-shelf ones were modular the way I needed. I designed it from scratch and built it by hand, which turned out to mean three days of thinking about tolerances.

I built one surface to aggregate my financial, physical, and operational data, because the context switching was adding up. I don’t use it anymore, but figuring out the API connections and the agentic AI architecture is what taught me I could build anything, and every project I’ve shipped since traces back to what I learned here. Sometimes the thing you build matters less than what it unlocks.

I lived near Stanford’s campus and wanted to capture how beautiful it actually is, frame by frame. I shot thousands of individual photos and taught myself the entire stitching and pacing process from scratch. The final cut lives on YouTube and scrapped more attempts than I’d like to admit.