🎓 About Me

I am a Ph.D. student with Prof. Xiao HUANG (since May 2025) at the Department of Computing, The Hong Kong Polytechnic University. Prior to that, I obtained my B.E. in Energy and Power Engineering from Huazhong University of Science and Technology (2020–2024).

My research interests include Retrieval-Augmented Generation (RAG), LLM Agents, and Tool Learning, with work published at ICLR and WWW. Currently, I focus on LLM agents, with a growing interest in embodied AI and spatial intelligence.

🎉 News

  • [2026.09] Our preprint Org-Agent is released! [Paper]
  • [2026.04] Our work LoSemB is published at WWW’26. [Paper]
  • [2026.02] Our survey on graph-based agent memory is released! [Paper|Github awesome-list] GitHub stars for Awesome-GraphMemory
  • [2026.01] Our work LinearRAG is accepted to ICLR’26. [Paper|Code] GitHub stars for LinearRAG

đź“” Selected Publications (Full List)

Preprint:

Published:

🏆 Honors and Awards

  • [2024] Pacesetter Merit Student, Huazhong University of Science and Technology
  • [2023] National Scholarship for Undergraduate Students

đź’Ľ Academic Service

  • Program Committee: WSDM (2026); ICLR (2026)

Outreach

I am happy to offer a 30-minute chat to early-stage Ph.D., master's, and undergraduate students, as well as potential collaborators. I especially welcome students from other disciplines who are interested in transitioning into AI research. Topics include LLM agents, retrieval-augmented generation (RAG), and agent memory, as well as research ideas, graduate school applications, and navigating research life. I am also interested in exchanging ideas on embodied AI and spatial intelligence. If you would like to chat, please fill out this form to request a meeting. Please feel free to send a follow-up email if you have not received a reply.

đź§© Projects

  • K-Cube: A knowledge graph-based platform for interactive teaching and learning. [Website]
  • DeepRead: An AI-powered research companion for paper organization and efficient reading. [GitHub|Download] GitHub stars for DeepRead