
Di Liu (刘迪)
Ph.D. Candidate
Department of Computer Science and Engineering
Shanghai Jiao Tong University (SJTU)
Email: liu-di AT sjtu DOT edu DOT cn
Github: diliu0349
Here is my Curriculum Vitae.
Ph.D. Candidate
Department of Computer Science and Engineering
Shanghai Jiao Tong University (SJTU)
Email: liu-di AT sjtu DOT edu DOT cn
Github: diliu0349
Here is my Curriculum Vitae.

Welcome to Di Liu's Homepage!
I am a first-year Ph.D. candidate at Shanghai Jiao Tong University (SJTU), fortunate to be advised by Prof. Chen Chen. Previously, I obtained my master's degree and my bachelor's degree from Beihang University (BUAA) and Beijing Jiaotong University (BJTU) in 2024 and 2021, respectively. During my first year as a Ph.D. student, I had a wonderful time working with Zhenhua Han, Huqiang Jiang, and Yuqing Yang from System Research Group at Microsoft Research Asia (MSRA).
[News]
[Research]
[Publications]
[Internship]
News
Oct 2024
RetrievalAttention accepted to NeurIPS 2024 ENLSP Workshop (Oral), Congratulations!
Jun 2024
Join the CSE department of SJTU as a Ph.D. student
Feb 2024
Join the System Research Group of MSRA as a research intern
Research
My research interest lies in computer systems, particular in machine learing systems.- Resource management and scheduling - to optimize resource utilization under various targets, including QoS and energy consumption (Morak).
- System and algorithn co-design - to optimize LLM inference by dynamic sparse attention (RetrievalAttention) and KV cache compression.
Publications
Internship

Microsoft Research Asia | System Research Group | Research Intern
Feb 2024 - Oct 2024 Mentor: Zhenhua Han, Huqiang Jiang, and Yuqing Yang.
Topics: Long-context LLM inference acceleration via dynamic and sparse attention computation.
Feb 2024 - Oct 2024 Mentor: Zhenhua Han, Huqiang Jiang, and Yuqing Yang.
Topics: Long-context LLM inference acceleration via dynamic and sparse attention computation.