Projects

AI-assisted query rewriting, program synthesis, and autonomous cloud-warehouse optimization.

AI systems that can test, diagnose, and demonstrate when their work is actually complete—and preserve consistent assumptions across durable workflows.

From bounded approximation to portable, interactive analytics over massive data.

Databases that get smarter with every workload.

Diagnosing, predicting, and explaining database performance from workload behavior.

Sampling, optimization, communication, and transformer methods for efficient machine learning.

Making database behavior, latency, resource use, and uncertainty more predictable.

Physical database designs that remain effective when workloads and estimates change.

From transaction-latency diagnosis to contention-aware scheduling adopted in major database engines.

Using active learning to make large-scale crowdsourced data acquisition more efficient.

Query languages and systems for mining streams and matching complex sequential events.