Zhiwei Liu
Member of Technical Staff · Recursive
I am a Member of Technical Staff at Recursive. Previously, I was a Senior Applied Scientist at Microsoft AI, where I worked on Bing Search, and a Senior Research Scientist at Salesforce AI Research from 2022 to 2026. I received my Ph.D. from the University of Illinois Chicago in 2022, advised by Philip S. Yu. I completed my undergraduate studies at the Yingcai Honor School at the University of Electronic Science and Technology of China (UESTC).
My current research focuses on Recursive Self-Improvement (RSI), Agent Data & Evaluation, and Agentic Information Retrieval. I have authored more than 100 papers in venues including NeurIPS, ICLR, KDD, SIGIR, WSDM, WWW, CIKM, and EMNLP, and hold 12 patents globally.
I also contribute to the research community as an Associate Editor, guest editor, workshop organizer, and program committee member across AI, natural language processing, information retrieval, data mining, and recommender systems.
News
- Two papers accepted to the NeurIPS 2026 Evaluations & Datasets Track. Topics: long-horizon healthcare agent workflows and real-world user decision modeling.
- I joined Recursive as a Member of Technical Staff.
- Two papers accepted to EMNLP 2026.
- I joined Bing Search at Microsoft AI as a Senior Applied Scientist.
Research interests
- Recursive Self-Improvement (RSI)
- Agent Data & Evaluation
- Agentic Information Retrieval
Selected publications
View more publications →- χ-Bench: Can AI Agents Automate End-to-End, Long-Horizon, Policy-Rich Healthcare Workflows?Haolin Chen, Deon Metelski, Leon Qi, Tao Xia, Joonyul Lee, et al.Conference on Neural Information Processing Systems (NeurIPS), Evaluations & Datasets Track, 2026.
- BehaviorBench: Modeling Real-World User Decisions from Behavioral TracesLiangwei Yang, Jielin Qiu, Zixiang Chen, Ming Zhu, Juntao Tan, Zhiwei Liu, et al.Conference on Neural Information Processing Systems (NeurIPS), Evaluations & Datasets Track, 2026.
- Test-Time Adaptation for LLM Agents via Environment InteractionA. Chen, Z. Liu, J. Zhang, A. Prabhakar, Z. Liu, S. Heinecke, S. Savarese, et al.International Conference on Learning Representations (ICLR), 2026.
- WebScale-RL: Automated Data Pipeline for Scaling RL Data to Pretraining LevelsZ. Cen, H. Chen, S. Wang, Z. Liu, Z. Liu, D. Zhao, C. Xiong, H. Wang, W. YaoInternational Conference on Learning Representations (ICLR), 2026.
- UserBench: An Interactive Gym Environment for User-Centric AgentsCheng Qian, Zuxin Liu, Akshara Prabhakar, Zhiwei Liu, et al.Conference on Empirical Methods in Natural Language Processing (EMNLP), 2026.
- UserRL: A Gym-Based Testbed for User-Centric Reinforcement LearningCheng Qian, Zuxin Liu, Akshara Prabhakar, Jielin Qiu, Zhiwei Liu, et al.Conference on Empirical Methods in Natural Language Processing (EMNLP), 2026.
- MCPEval: Automatic MCP-based Deep Evaluation for AI Agent ModelsZhiwei Liu, Jielin Qiu, Shiyu Wang, Jianguo Zhang, et al.arXiv preprint, 2025.
- xLAM: A Family of Large Action Models to Empower AI Agent SystemsJianguo Zhang, Tian Lan, Ming Zhu, Zuxin Liu, et al.arXiv preprint, 2024.
- PRACT: Optimizing Principled Reasoning and Acting of LLM AgentZhiwei Liu, Weiran Yao, Jianguo Zhang, Rithesh Murthy, et al.Conference on Natural Language Learning (CoNLL), 2024.
- BOLAA: Benchmarking and Orchestrating LLM-augmented Autonomous AgentsZhiwei Liu, Weiran Yao, Jianguo Zhang, Le Xue, et al.LLMAgent Workshop at ICLR, 2024.
- DialogStudio: Towards Richest and Most Diverse Unified Dataset Collection for Conversational AIJianguo Zhang, Kun Qian, Zhiwei Liu, Shelby Heinecke, et al.arXiv preprint, 2023.
For a complete and current list, see my Google Scholar profile.
Professional service
View full service & activities →Associate Editor of the International Journal of Machine Learning and Cybernetics; Guest Editor for ACM TORS and Electronics; co-chair and organizer of Agent4IR workshops at KDD and CIKM; and program committee member for NeurIPS, CIKM, WWW, RecSys, AAAI, IJCAI, ACL, NAACL, EACL, KDD, SDM, and ECML-PKDD.