Welcome to Zeju Li (李泽钜) ‘s homepage!

I’m a 3rd-year PhD Candidate in The Chinese Univeristy of Hong Kong, the Department of Computer Science and Engineering, supervised by Prof. Qiang Xu.

Before that, I got my bachelor’s degree in ‘class for talent’ of Beijing University of Posts and Telecommunications (BUPT) with honor. I also have some wonderful research experiences in Institute of Automation in CAS, Institute of Computing Technology in CAS, Bytedance.

My research interests include Post-Training (RL) for Foundation Model, Long-term Agent Memory, LLM Efficiency and VLM.

I am in 2027 job market and actively looking for academic collaboration, feel free to drop me an email. (lizeju0727@gmail.com)

Education

Beijing University of Posts and Telecommunications, 2020.09 - 2024.06

B.E. in School of Information and Communication Engineering, class for talent

The Chinese Univeristy of Hong Kong, 2024.08 - 2027.11 (expected)

Ph.D in Department of Computer Science and Engineering

News

• [2026.07] Our paper Latent Context Compilation: Distilling Long Context into Compact Portable Memory was accepted by COLM 2026. See you in San Francisco🇺🇸.
• [2026.04] I got Gold Reviewer Award. See you in Seoul🇰🇷.
• [2026.04] Our project for OpenClaw Security Solution is released: https://arbiteros.ai/
• [2026.04] Our paper Solve-Detect-Verify: Inference-Time Scaling with Flexible Generative Verifier was accepted by ACL 2026 Main Conference. See you in USA🇺🇸.
• [2026.01] Our paper Reasoning Scaffolding: Distilling the Flow of Thought from LLMs was accepted by ICLR 2026. See you in Rio🇧🇷.
• [2026.01] Our paper Making Slow Thinking Faster: Compressing LLM Chain-of-Thought via Step Entropy was accepted by ICLR 2026. See you in Rio🇧🇷.
• [2025.09] Our paper Dependency Matters (LLM Reasoning) was accepted by NeurIPS 2025. See you in USA🇺🇸.
• [2025.08] Our paper ThinkAnswer Loss was accepted by EMNLP 2025 Findings. See you in Suzhou🇨🇳.
• [2025.08] Our paper Dyve was accepted by EMNLP 2025 Main Conference. See you in Suzhou🇨🇳.
• [2025.06] Our paper Argus was accepted by TNNLS.
• [2025.05] Our paper DeepCircuitX was accepted by ICLAD 2025 as Oral. See you in USA🇺🇸.
• [2025.04] Our paper ForgeEDA was accepted by ISEDA 2025 as Oral.
• [2025.03] Our paper Spatial 3D-LLM was accepted by ICME 2025 as Oral.
• [2025.01] Our paper DeepRTL was accepted by ICLR 2025, and will be presented as a Spotlight.
• [2024.11] We got Honorable Mention Award of LLM4HWDesign Contest of ICCAD 2024.
• [2024.08] Fortunately got Huawei PhD Fellowship.
• [2024.05] I join Cure Lab in CUHK as research assistant, supervised by Prof.Qiang Xu.
• [2024.04] I join KAUST as a visiting student, supervised by Prof.Mohamed Elhoseiny.
• [2024.04] Our work 3DMIT is accepted by ICME-3DMM 2024 and selected as candidates of the Best Paper Award.
• [2023.10] I receive Singapore International Pre-Graduate Award(SIPGA) by I2R of A STAR.
• [2023.08] I join Beijing Shuyuan Research Center as a research intern. My research interets include multi-modal LLM and 3D world LLM.
• [2023.07] I join The Chinese University of Hong Kong(Shenzhen) by Prof.Xiaoguang Han. My research interets and fields are AIGC and 3D generation.
• [2023.04] Our work about “knowledge graph construction and fusion” is accepted “ICALT 2023” and receives the Best Paper Award Nomination.
• [2023.04] I work at the National Laboratory of Pattern Recognition (NLPR) at the Institute of Automation of the Chinese Academy of Sciences (CASIA) as a research intern. My research interets and fields are diffusion model and AI4Science.
• [2023.01] Our work on “robust and safety about semantic communication” is accepted by “ICC 2023”and “IEEE Communciation Magazine”.
• [2022.08] Our work on “adversarial attacks for semantic communication” is accepted by “2022 IEEE 96th Vehicular Technology Conference”.
• [2022.05] I join in the National Engineering Research Center of Mobile Network Security in BUPT as a research intern.
• [2021.05] I join in the State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecomunications.