ShadeFinder
Urban shadow mapping to find cooler walking routes.
A geospatial project that identifies shaded walking paths through a city, so you can pick the cooler route rather than the shortest one.
Source is currently private. Happy to walk through it on request.
How it works
Take building footprints and heights, take the sun's position for a given time and date, project the shadows, and then treat "shade" as a cost on a routing graph. The shortest path and the coolest path are rarely the same path.
Why I like this one
It's the most physically grounded thing I've built. There's no model risk, no distributional assumption to get wrong — the sun is where it is, buildings are where they are, and the geometry either checks out against reality or it doesn't. You can walk outside and falsify it.
That's a rare luxury. In Optimal Execution you can never fully observe the counterfactual: you don't get to see what the price would have done had you not traded. Here you just look up.
Stack: Python, geospatial analysis, Streamlit.
Related: Things I've Built