About Zhanli Li (李展利)

👋 Hi, I’m Zhanli Li (李展利), an incoming PhD student at the LAMDA led by Prof. Zhi-Hua Zhou (周志华), Nanjing University, where I will be advised by Prof. Han-Jia Ye (叶翰嘉). I am currently pursuing a bachelor’s degree in Digital Economy (2023-2027) at the Wenlan School of Business, Zhongnan University of Economics and Law (ZUEL). I am currently interning at LongCat Interaction, focusing on the post-training scaling of working coding agents. During my time at ZUEL, I’ve had the privilege of meeting an amazing group of senior students with incredible insights into deep learning. As of now, my open-source projects have earned and my publications as first author have received . My first principle in research is to make AI research that helps change the world for the better, and I am currently deeply interested in agentic training, document intelligence, data intelligence, and potential explainability.

🎓 Previously, from 2024 to February 2025, I was mentored by Prof. Zichao Yang (杨子超) at ZUEL, working on causal inference. From February 2025 to March 2026, I was an intern at ICT and PAI TECH (庖丁科技), supervised by Prof. Yixuan Cao (曹逸轩), working on document intelligence with Prof. Ping Luo (罗平)’s team.

💬 Beyond academic research, I enjoy playing table tennis, running, and cycling in my free time. I am also a big fan of video games, especially Valorant—where I mainly play as a Duelist, frequently locking in Phoenix, Neon, and Omen—as well as MOBA mobile games. I welcome anyone interested in my work or seeking potential collaborations to reach out via email: lizhanli@stu.zuel.edu.cn.

Selected Publications 📓

Before introducing my selected publications, I would like to share a few thoughts on how I think about research. In the future, I hope to lead only one paper each year. For each work, I want to start from a problem that I find both interesting and meaningful, spend time understanding its principles and foundations, and then work on it seriously. If the problem is solved, that would be wonderful—a piece of good work. If not, that is also fine; it may still lead to an incremental paper. What truly matters is my own growth and the fact that I have helped push the boundary of the problem forward.

• Zhanli Li, Huiwen Tian, Lvzhou Luo, Yixuan Cao, Ping Luo.
  DeepRead: Document Structure-Aware Reasoning to Enhance Agentic Search
  New Intelligence (新智元)
  
• Zhanli Li, Yixuan Cao, Lvzhou Luo, Ping Luo.
  Navigating Large-Scale Document Collections: MuDABench for Multi-Document Analytical QA Findings of ACL 2026
  Hugging Face Dataset
  
• Zhanli Li, & Zichao Yang.
  ESG rating disagreement and corporate Total Factor Productivity: Inference and prediction Finance Research Letters (JCR Q1, IF 6.9)
  

News 📧

• 2026.6.16 We ranked 119/4182 and earned a Silver Medal in the NVIDIA Nemotron Model Reasoning Challenge on Kaggle.
• 2026.5.17 One paper about Agentic Search was invited to Resubmit to KDD 2027 (First Cycle).
• 2026.4.6 MuDABench was accepted as Findings of ACL 2026!
• 2026.3.17 DeepRead was covered by New Intelligence (新智元).
• 2026.2.9 One paper about Agentic Search was submitted to KDD26.
• 2025.10.12 I am honored to have received the China National Scholarship (Top 0.2% in china)!
• 2025.9 I lead an open-source project for automated outreach, it’s Auto-Tutor!
• 2025.8.1 ESG rating disagreement and corporate Total Factor Productivity: Inference and prediction was awarded an outstanding paper by Tsinghua University (Top 3% in all participants)!
• 2025.3.13 I was invited as a reviewer of Finance Research Letters.
• 2025.3.8 ESG rating disagreement and corporate Total Factor Productivity: Inference and prediction had been published online.
• 2025.2 I joined the Key Laboratory of Intelligent Information Processing, Institute of Computing Technology, Chinese Academy of Sciences and will work with Prof. Yixuan Cao.

Vistor 🗺️