About Me
I am a Ph.D. student in the Gaoling School of Artificial Intelligence at Renmin University of China, advised by Prof. Xu Chen. I am a member of the AI-Engine Lab. I joined the integrated master's–Ph.D. program in 2022 and expect to graduate in 2028. Previously, I received my bachelor's degree in Computer Science and Technology from Wuhan University of Technology in 2022.
My research focuses on LLM-powered recommender systems and personalized agents. I study reasoning for recommendation, user intent understanding, and interactive agents with long-term memory. I am also interested in faithful user simulation and evaluation, connecting algorithmic advances with real-world applications.
I have worked on industrial recommendation at Taobao through the RecGPT series (V1 · V2 · V3), and on reasoning for industrial ranking through OnePiece. My recent research interests include efficient long-sequence modeling for large language models, personalized memory, and personalized agents.
Outside of research, I enjoy basketball, running and other sports, as well as movies and music. A long-term dream of mine is to build a one-person company someday.
Citation metrics from Google Scholar, updated Oct 26, 2026
Selected Publications
-
RecGPT-V3 Technical Report
A stateful, hybrid-modal LLM recommender that reasons over text and Semantic IDs, with a Memory Hub for long-term user memory and latent intent reasoning. In Taobao's homepage feed it lifts GMV by 3.97% while cutting serving cost by 52.4%.
First authorPaper
-
RecGPT-V2 Technical Report
Restructures intent reasoning as a hierarchical multi-agent system, with meta-prompted explanations, constrained RL and an Agent-as-a-Judge evaluator. Cuts GPU use by 60% and lifts CTR by 2.98% in Taobao A/B tests.
First authorPaper
-
RecGPT: A User Intent-Centric Next-Generation LLM-Powered Recommender System in Industrial Practice
Puts user intent at the center of the recommendation pipeline, using LLMs for interest mining, item retrieval and explanation generation. Fully deployed on the Taobao app.
First authorPaper
-
OnePiece: Bringing Context Engineering and Reasoning to Industrial Cascade Ranking System
Brings LLM-style context engineering and block-wise latent reasoning into industrial retrieval and ranking models. Deployed in Shopee's main search, with over 2% GMV/UU gain.
Co-first authorPaper
-
Think Before Recommend: Unleashing the Latent Reasoning Power for Sequential Recommendation
The first inference-time computing framework for recommendation. Implicit multi-step reasoning raises the performance ceiling of sequential recommenders by roughly 30% to 50%.
-
Interactive recommendation agent with active user commands
Lets users steer the recommendation feed with natural-language commands, through a Parser agent and a Planner agent. Improves both user satisfaction and business metrics in long-term online tests.
First authorPaper
-
Towards Faithful Simulation of Human Shopping Behavior
A GUI-grounded agent that simulates real shopping sessions from screenshots, using hierarchical memory and trajectory-level RL. Also releases the USB benchmark.
First authorPaper
-
LoopCTR: Unlocking the Loop Scaling Power for Click-Through Rate Prediction
Scales CTR models by looping shared layers instead of adding parameters. Trained with multiple loops, it beats all baselines even with zero loops at inference.
First authorPaper
News
- RecGPT-V3 received the 🏆 SIGIR-AP GRIT Workshop Best Paper award.
- Received my fourth National Scholarship, following awards in 2019, 2020, and 2021.
- Released UserArena (co-first author; proposed the core idea and co-wrote the paper), a benchmark for interactive user simulation with long-horizon shopping agents.
- Released LoopRec (co-first author; proposed the core idea and co-wrote the paper), exploring item-level adaptive recursion for sequential recommendation.
- ReaRec (first author) was published in IEEE TKDE.
- Released RecVerse (first author), a GUI-grounded shopping simulation agent with hierarchical memory and trajectory-level reinforcement learning.
- OnePiece (co-first author) was published at KDD 2026 with an 🌟 Oral presentation.
- RecBot (first author) was published at KDD 2026.
- OneRank (first author) was published at KDD 2026.
- RecGPT (first author) was accepted by ACM TOIS.
Internships
- 2025.04–Present
Alibaba · TaotianT-StarTalent Intern
Recommendation Algorithm Intern
- 2024.11–2025.03
Tencent · WeChat
LLM Agent Algorithm Intern
- 2024.05–2024.11
Alibaba · Tongyi
LLM Agent Algorithm Intern
- 2023.10–2024.04
Kuaishou
Recommendation Algorithm Intern
- 2023.02–2023.07
Huawei · Noah's Ark Lab
Recommendation Algorithm Intern
Education
- 2022.09–Present
Renmin University of China
Integrated Master's–Ph.D. Program in Artificial Intelligence
Gaoling School of Artificial Intelligence · Advisor: Xu Chen
Expected graduation: 2028 - 2026.05–2026.11
National University of Singapore
Joint Ph.D. Training · School of Computing
NExT++ Lab
Advisors: See-Kiong Ng and Yang Zhang - 2018.09–2022.06
Wuhan University of Technology
Bachelor's Degree in Computer Science and Technology
Ranked 1 / 257
Honors and Awards
Research Awards
- 2026.12SIGIR-AP GRIT Workshop Best PaperRecGPT-V3 Technical Report · First author
- 2022.10CIKM Best Resource Paper Runner-UpRecBole 2.0 · Co-author
Scholarships & University Honors
- 2026.12National Scholarship博士研究生国家奖学金Renmin University of China · Ph.D.
- 2019–2021National Scholarship本科生国家奖学金Wuhan University of Technology · UndergraduateThree-time recipient · 2021.12, 2020.12, 2019.12
More university honors (8)
- 2023–2025Academic Excellence Scholarship学习优秀奖学金Renmin University of ChinaFirst class: 2024.12 · Third class: 2025.11, 2023.11
- 2023–2025Merit Student三好学生Renmin University of ChinaTwo-time recipient · 2025.09, 2023.09
- 2023.09Outstanding Student Leader优秀学生干部Renmin University of China
- 2023.09Third-Class Scholarship for Social Work and Volunteer Service社会工作与志愿服务骨干三等奖学金Renmin University of China
- 2022Outstanding Graduate优秀本科毕业生Wuhan University of Technology
- 2019–2020Model Merit Student校三好学生标兵Wuhan University of TechnologyTwo-time recipient · 2020.11, 2019.11
- 2020Outstanding Youth League Cadre校优秀共青团干部Wuhan University of Technology · Academic year 2019–2020
- 2019.05Outstanding Youth League Member校优秀团员Wuhan University of Technology · Academic year 2018–2019
Competition Awards
- 2021.05Silver Award, National Finals中国高校计算机大赛团体程序设计天梯赛全国总决赛银奖China Collegiate Computing Contest · Group Programming Ladder Tournament
- 2021.02First Prize, Programming Finals全国高校计算机能力挑战赛程序设计赛决赛一等奖National College Computer Ability Challenge
Invited Talks
Reasoning-Enhanced Recommendation
Kuaishou · Beijing Yuan Center (北京元中心)
推理增强推荐
A General Social Simulation Platform with Large Language Model-Based Agents
CCF YOCSEF · Jinan
基于大模型智能体的通用社会模拟平台
Academic Services
Workshop Organization
- 2027
HCRS @ WWW 2027
Core Organizer3rd Workshop on Human-Centered Recommender Systems
- 2026
HCRS @ WWW 2026
Core Organizer2nd Workshop on Human-Centered Recommender Systems
Conference Volunteering
- 2024
KDD 2024
Student Volunteer
Open-Source Software
Fairness-aware recommendation toolkit in the RecBole ecosystem.
Contributions to the unified recommendation framework.
Contributions to core functionality and evaluation.
Conference & Journal Service
Conferences
Reviewer- NeurIPS
- ICLR
- AAAI
- IJCAI
- KDD
- SIGIR
- WSDM
- ACL
- EMNLP
- WWW
Journals
Reviewer- ACM TOIS
- IEEE TKDE
- ACM TORS