About me

I am a second-year Master’s student at Tsinghua University. My research mainly lies in Large Language Models (LLMs), Agentic Reinforcement Learning (Agentic RL), and Information Retrieval (IR), with a recent emphasis on Web Agents.

A core motivation of my work is to use interaction with real environments to improve the base competence of models, such as their planning ability, robustness, and generalization. Ultimately, I aim to build more capable and reliable foundation models.

I am currently seeking PhD positions starting in Fall 2027, with a primary focus on opportunities in mainland China, Hong Kong, and Singapore.

Research Interests

My current and past research topics include:

• LLMs & Agentic Reinforcement Learning
  • Building web agents that autonomously browse, understand, and operate complex websites to complete long-horizon tasks.
  • Using web interaction as a training ground to enhance the fundamental capabilities of models, such as:
    • Strategic planning and task decomposition,
    • Tool use and API interaction,
    • Handling noisy, dynamic, and partially observable environments.
  • Integrating RL with LLMs so that agents can learn from trial-and-error in realistic tasks, instead of relying only on offline supervised data.
  • Studying how agentic interaction can feed back into better pretrained/backbone models (e.g., via behavior data, trajectories, and preferences).
• Information Retrieval
  • Retrieval-Augmented Generation (RAG)
    • Designing retrieval and reranking pipelines to support LLMs with accurate and up-to-date external knowledge.
    • Analyzing how retrieval quality and system design affect downstream reasoning and generation robustness.
  • Recommender Systems (Recsys)
    • Exploring retrieval and ranking methods in recommender systems, especially under sparse signals or long-term user feedback.
    • Studying how interaction data and user behavior can be exploited to improve both retrieval and personalized recommendation.

Selected Papers (First-Author)

• What Should I Cite? A RAG Benchmark for Academic Citation Prediction
  Leqi Zheng*, Jiajun Zhang*, Canzhi Chen*, Chaokun Wang, Hongwei Li, Yuying Li, Yaoxin Mao, Shannan Yan, Zixin Song, Zhiyuan Feng, Zhaolu Kang, Zirong Chen, Hang Zhang, Qiang Liu, Liang Wang, Ziyang Liu
  The Web Conf 2026
  [ Paper | Code ]

• Negative Feedback Really Matters: Signed Dual-Channel Graph Contrastive Learning Framework for Recommendation
  Leqi Zheng, Chaokun Wang, Zixin Song, Cheng Wu, Shannan Yan, Jiajun Zhang, Ziyang Liu
  NeurIPS 2025
  [ Paper | Code ]

• LAGCL4Rec: When LLMs Activate Interactions Potential in Graph Contrastive Learning for Recommendation
  Leqi Zheng, Chaokun Wang, Canzhi Chen, Jiajun Zhang, Cheng Wu, Zixin Song, Shannan Yan, Ziyang Liu, Hongwei Li
  EMNLP 2025
  [ Paper | Code ]

• Balancing Self-Presentation and Self-Hiding for Exposure-Aware Recommendation Based on Graph Contrastive Learning
  Leqi Zheng, Chaokun Wang, Ziyang Liu, Canzhi Chen, Cheng Wu, Hongwei Li
  SIGIR 2025
  [ Paper | Code ]

• Audio-Visual World Models: Towards Multisensory Imagination in Sight and Sound
  Jiahua Wang*, Shannan Yan*, Leqi Zheng*, Jialong Wu, Yaoxin Mao
  Under review
  [ Paper]

Honors and Awards

• 2024   Excellent Higher Education Graduate of Beijing Municipality
• 2023   China National Scholarship
• 2022   China National Scholarship

Experience

• Research Intern, THUIR, Department of Computer Science and Technology, Tsinghua University, 2025
  • Work on Information Retrieval.
• Research Intern, Ant Group, 2025
  • Work on Agentic RL and LLM post-training.
• Research Intern, JD.com, Inc., 2024
  • Work on Data Mining and Recommender Systems.
• Research Intern, Wangxuan Institute of Computer Technology, Peking University, 2023
  • Work on Recommender Systems.

Last updated: Jan. 2026