I’m leading the engineering strategy to build a specialized ML platform for onchain social. We’re constructing a full-cycle pipeline ingestion, feature store, training, and serving, designed to handle the unique sparsity and velocity of blockchain data.

Our first target is a personalized discovery feed powered by a multi-stage ranker. It combines traditional social signals (NLP-based feature extraction, safety detection) with real-time onchain activity to surface high-value interactions. It’s been a fun challenge architecting the foundation for a decentralized social graph that feels as responsive and relevant as a centralized one.

Working to deliver a social and creator experience on the base app, Coinbase’s new consumer super app.