Research Summary
The goal of my research is to make warehouse-scale parallel systems efficient so that big data applications (e.g., DNN training, analytics) can run 100–1000x faster and 10-100x cheaper. For that goal, I have three thrusts: (1) Flash burst computing (2) Resource efficient deep learning (3) Distributed systems for energy efficiency.
To learn more about my research, please visit my
website.
I previously worked on many areas in distributed systems: distributed deep learning system, distributed analytics, blockchain, resource fungibility, low-latency consensus, distributed system consistency, in-memory large-scale storage, server overload control. I usually publish to networked systems venues, such as NSDI, OSDI, SOSP.
