· Developed and launched a recommendation platform, leading cross-functional efforts that successfully integrated personalized recommendations into 25 different applications on Cash App· Implemented a unified semantic search platform powered by LLMs and an embedding store, leading to a 45% increase in recall of relevant results across merchant and product search· Built low-latency machine learning models to power Cash's real-time P2P search systems, with the resulting integrations driving a 30% increase in recall, 15% increase in precision, and a 10% reduction in latency· Reduced search model development time from months to days for teams engineers and data scientists with a suite of tools to run controlled search experiments in a backtesting environment· Created batch and streaming feature pipelines to compute predictive features for online inference, fueling an increase in search and recommendation metrics