Scholar
Erkang Zhu
Google Scholar ID: 8vrVMwIAAAAJ
Microsoft Research, University of Toronto
Citations & Impact
Citations
3,816
H-index
17
i10-index
22
Publications
20
Co-authors
21
list available
Publications
Resume (English only)
Academic Achievements
- Lead developer of the open-source framework AutoGen
- Significant performance improvements in MATCH_RECOGNIZE query processing (5.4X and 6X)
- Developed algorithms for large-scale set similarity search and data sketches, enabling millisecond-level discovery of joinable/unionable tables among 100K+ tables
- Created an Open Data search engine stack
- Active academic and open-source presence via GitHub, Google Scholar, and personal blog
Research Experience
- Principal Researcher at Microsoft Research; lead architect of the open-source AI agent framework AutoGen
- AutoGen is based on the Actor Model, offering high-level and event-driven low-level APIs for building single and multi-agent applications
- Previously focused on query processing, developing a cost-based, platform-independent rewrite rule for MATCH_RECOGNIZE queries, achieving 5.4X median latency improvement in Trino
- Developed a specialized execution engine for MATCH_RECOGNIZE with extended operators and a novel cost-model-based optimizer, yielding 6X median performance gain over state-of-the-art engines
- Built an Open Data search engine stack based on PhD research to facilitate Open Data usage in applications
