Problem
Base network gas prices fluctuate significantly, and users often pay more than necessary for transactions. The challenge was to build a system that predicts gas prices and helps users time their transactions optimally to save on fees, with potential savings of up to 40%.
Approach
Built a full-stack ML application with React/TypeScript frontend and Python Flask backend. Used scikit-learn models to predict Base network gas prices based on real-time blockchain data, integrating with MetaMask for wallet support.
- ML model development and training
- Full-stack architecture (React + Flask)
- Real-time blockchain data integration
Implementation
Developed a system featuring live gas indicators, 24-hour price predictions, and a savings calculator. The application integrates real-time blockchain data to provide accurate predictions and helps users optimize transaction timing.
Results
Won the QMUL AI Society x Coinbase Hackathon. The application successfully predicts gas prices and helps users save up to 40% on transaction fees by timing their transactions optimally. The full-stack implementation demonstrates end-to-end ML system development.