At Airbnb, I tech-led a TensorFlow multi-task ranking model for Stays search, optimizing click, booking, and cancellation objectives and driving a significant lift in uncancelled bookings in global A/B tests.
I owned PyTorch two-tower retrieval for flexible-date search, adding availability-aware hard negative sampling that improved retrieval recall and replaced a hand-tuned geo filter. I also built an LLM-as-judge labeling pipeline and shipped a LangGraph agent that turns natural-language queries into structured search filters.
At BlackRock, I developed equity return and risk forecasting models and NLP pipelines that added sentiment and event signals to investment research. I’m grounded in data science, statistics, and finance, and have worked across machine learning, search relevance, recommendation systems, and LLM evaluation.
