Full-stack predictive analytics platform for NCAA basketball betting, delivering 56% prediction accuracy and 6% ROI across 300+ tracked predictions. Enables users to create custom models trained on 43,000+ games with 50 engineered features. Built with React/TypeScript frontend, async Python backend, and automated ETL pipelines. Fully deployed on AWS (Lightsail, EventBridge, CloudWatch) with CI/CD for continuous data ingestion and model updates.
