Reforming the Mechanism: Editing Reasoning Patterns in LLMs with Circuit Reshaping
Ph.D. Student · Capital One Fellow
Zhenyu (Fischer) Lei
E-mail: vjd5zr [at] virginia [dot] edu
Address: Charlottesville, VA, USA
About
About me
I am a Ph.D. student in Electrical and Computer Engineering at the University of Virginia, advised by Prof. Jundong Li. Before that, I received my B.S. in Physics from Xi'an Jiaotong University, where I was lucky to join the LUD Lab. I have also interned at AT&T as an Applied Scientist.
My research centers on LLM continual learning, with a broad interest in graph and time series. I am open to collaborations and discussions — feel free to reach out via email! Actively looking for intern!
News & notices
News
Sep 2026One paper accepted at NeurIPS 2026 (AdaST).
Aug 2026Two papers accepted at EMNLP 2026 (one Main (MemSuit), one Findings (TeacherGRPO)).
June 2026Honored to recieve Capital One Fellow
Jan 2026One paper accepted at ICLR 2026 (REdit).
Jan 2026One paper accepted at ISBI 2026 (BrainTAP).
Oct 2025One paper accepted at WSDM 2026 (MolEdit).
Publications
All publications →
Selected Research
30papers
11first-author papers
Grounding Memory Summarization in Utility Intent
AdaST: Adaptive Coupling for Spatial-Temporal Forecasting
* equal contribution
Background
Education & Experience
UVA
University of Virginia, USAPh.D. in Electrical and Computer Engineering · Advisor: Prof. Jundong Li
2023.08 – present
XJTU
Xi'an Jiaotong University, ChinaB.S. in Physics (Honor) · GPA 89.5/100 · Advisor: Prof. Minnan Luo
2019.08 – 2023.07
AT&T
AT&T, Bedminster, NJApplied Scientist Intern · Host: Dr. Qiong Wu
2025.06 – 2025.08
Recognition
Selected Honors & Awards
Capital One PhD Fellowship, 2025–26WSDM Student Travel AwardPresidential Fellowship in Collaborative Neuroscience, UVAPrecision Health Initiative Trainee Travel Stipend, UVATYDE Student Training Award, UVAEngineering Academic Commitment to Excellence (ACE) Fellowship, UVA
Community
Service
Reviewer: CIKM (2025), COLM (2024, 2025), KDD (2024, 2025), ARR (Dec 2023–), NeurIPS (2023)Volunteer: ACL 2023 (virtual)