
Parkeye
A parking-availability forecasting model that predicts campus lot occupancy up to 4 hours ahead. Built an ETL pipeline from 5 years of historical data (25,700+ records across 14 George Mason lots), trained an XGBoost quantile regression model, and shipped it as a FastAPI microservice powering a React Native app — piloted with George Mason parking staff after incorporating 400+ student feedback responses.





