About
Hi! I'm Zhipeng Xu, a Senior Researcher at Alibaba Group, where I build foundational platforms for large language models and work toward advancing agents in complex real-world production environments, bridging scientific innovation with practical impact.
My research focuses on Long-Horizon Agents and Recursive Self-Improvement (RSI) systems. I study how agents can make sustained progress on open-ended tasks, learn from experience, and improve their own learning and problem-solving processes. My goal is to build reliable, adaptable agents that can tackle complex real-world problems and accelerate scientific discovery.
News
Our survey, Recursive Self-Improvement in AI: A Survey, is now available!
We are pleased to announce that CC-OCR V2 has been accepted to KDD 2027. It is a frontier OCR benchmark that we designed to establish the evaluation for Qwen3.8-Max.
Three papers accepted to ACL 2026: UNIKIE-BENCH in the main conference; Lang2Act and Genii in Findings.
Selected Publications
View All →* Equal contribution† Project Leader
Cleaner Pretraining Corpus Curation with Neural Web Scraping
Zhipeng Xu, Zhenghao Liu, Yukun Yan, Zhiyuan Liu, Ge Yu, Chenyan Xiong
ACL
Mixture-of-Retrieval Experts for Reasoning-Guided Multimodal Knowledge Exploitation
Chunyi Peng*, Zhipeng Xu*, Zhenghao Liu, Yishan Li, Yukun Yan, Shuo Wang, Yu Gu, Minghe Yu, Ge Yu, Maosong Sun
SIGIR