Our group’s research focus is on the implementation (as opposed to
applications) of data systems. Broadly, our group works on machine programming
for data systems, investigating how to build data systems that automatically
adapt to new hardware and user workloads, invent novel processing strategies,
and understand user intention.
Active projects
My group's projects include:
Offline query optimization: important queries deserve additional optimization. How can we find the best possible query plan given a time budget?
Scalable semantic operators: how can we scale semantic operators to datasets with billions of rows? Even calling a small LLM per-row is too slow.
Rethinking execution engines: as hardware and query workloads become more complex, how should we design exeution engines for future DBMSes?
LLMs for query optimization: LLMs have great reasoning capabilities in some domains. How can we leverage those abilities for QO?
Databases in the wild: comparing data management reality with assumptions made by academics and industry practitioners.
I also collaborate on:
TIDES: How can organizations share data while ensuring compliance with privacy regulations? TIDES is a privacy-preserving cross-organization data integration service.
TruthTable: how can someone verify that an external database is giving them the correct answer? TruthTable is a verifiable query engine that builds proofs that a result is correct.
Are you interested in working with us? Fill out this application. Note that we generally work with students who have strong systems backgrounds.
