A selection of things I've built. More coming soon.
Exercise in causal inference to compare the bikes I have ridden using strava api. Goal is to quantify how much faster a new bike can make you. Isolating the impact of the bike itself. Also looking to compare bike to bike accross different disciplines ie sprint, climb, flat. Basically the poor mans wind tunnel.
Data dashboard comparing my Seattle to Portland ride across two years — analyzing pacing, power, and splits. Generated entirely via MCP: I prompted Claude with Strava MCP access and it built the full dashboard autonomously. I can't take credit for this, as it was literally a 1 shot prompt, but I think its rpetty cool.
A planned calorie-tracking calibration project comparing calories burned, calories consumed, and weight change over the past couple of months to estimate how accurately my tracking reflects the expected 3,500-calorie-per-pound relationship.
Once developed this project will optimize fuel for a race. I plan to include inputs such as course, planned watt output, desired carbs/electrolytes per hour.
A planned descent optimization project that combines streaming ride data, gradient-driven physics, and AI training to model the fastest possible line down a local descent I ride often called The Snake.
GH Action driven AI pipeline that created custom shirts for every game of the March Madness basketball tournamentand posted them for sale on shopify. (deactivated)
This app is a UI to play with a stoplight animation. Unless you lnow the secret word in which case it controls a real stoplight that I obtained.