End-to-end owned, designed, developed, and launched the Uber Eats home feed dish recommendation carousel to 90M global users, boosting top-level business metrics. Coded in Java, Go, PySpark, and HiveQL.

Trained and indexed DL embeddings in Uber’s homegrown search system. Served embeddings for candidates retrieval using a novel approach that elevated recall rate by 4x with the same resource as baseline.

Implemented eater history retrieval based on personalized order and click data.

Prepared feature pipelines. Trained, tuned, and served a XGBoost model for candidates ranking.