AI for Tabular Data Discovery and Analytics

Develops intelligent methods for discovering, completing, and querying tabular data across large-scale data lakes. This line of work covers tabular data discovery, missing value imputation, and numerical table question answering, enabling machines to locate relevant tables from heterogeneous data sources, recover incomplete data using hybrid structural and semantic signals, and answer complex numerical questions over semi-structured and multi-table settings — all under practical cost and efficiency constraints.