Meta’s GPU-native retrieval and ranking stack

Database and Storage Lab in Alibaba DAMO Academy. Focus on Database Performance Optimization, Database Tuning and Database Diagnose boosted by AI models. Check out part of my published work:

• An Efficient Transfer Learning Based Configuration Adviser for Database Tuning (VLDB 2024)

• A Unified and Efficient Coordinating Framework for Autonomous DBMS Tuning (SIGMOD 2023)

• Towards Dynamic and Safe Configuration Tuning for Cloud Databases (SIGMOD 2022)

• ResTune: Resource Oriented Tuning Boosted by Meta-Learning for Cloud Databases (SIGMOD 2021)

• Leaper: A Learned Prefetcher for Cache Invalidation in LSM-tree based Storage Engines (VLDB 2020)

Petuum is one of the top startups in the first wave of AI. I led a team building a AI product on the cloud called Symphony. The backend is our in-house developed distributed training/inference infra, leveraging parameter servers, SSP and clippers. I was also contributing to the distributed training forked from TensorFlow v0.8, Serving, DyNet framework(which is earlier than PyTorch).

DB kernel developer, worked on the Open-Source Apache project HAWQ: distributed SQL Engine on Hadoop.

• Design and implement the HAWQ register feature

• Design and implement external ORC format support, FDW feature for HAWQ

• Implement the new feature test framework for HAWQ

• Top #3 committer