FanCall Fast Coding Hackathon
A build-mode hackathon at ZoHouse where I designed a hybrid recommendation system for Fancall, blending rule-based and collaborative filtering with a Python, FastAPI, and SQLite backend.
ZoHouse in the middle of build season is the closest I have felt to a writer's room for software. That day, the brief was Fancall: move the app from search to discovery, and make a recommendation system that actually understands people.
We did not build something generic. The ranking weighted category fit most heavily, then past interaction, then popularity, with location as the deciding edge, and it handled brand-new users through a smart-defaults path so nobody hit a cold empty screen. I built the whole thing in Python with FastAPI and SQLite, and shipped both the code and live API docs.
It was just one afternoon of deep work, but it taught me something about ranking that stays with me: the best systems feel obvious because the thought went into the edges, not the middle.
