BAAI CONFERENCE

Liu Weiwen, Associate Professor at Shanghai Jiao Tong University, is currently an Associate Professor and Doctoral Supervisor at the School of Computer Science, Shanghai Jiao Tong University. She once worked as a Principal Researcher at Huawei Noah's Ark Lab and received her Ph.D. degree from The Chinese University of Hong Kong in 2020. Her research mainly focuses on large model agents, large language models and information retrieval. She has published over 80 papers in top-tier international conferences and journals, holds 12 research patents, and has one ESI Highly Cited Paper. She has received numerous honors including the DLP-RecSys Best Paper Award, the Special Prize for Top 10 Vertical Domain Large Models, Huawei 2012 CEO Individual Award and the Innovation Pioneer Award. The open-source ToolACE series models developed by her team rank first among counterparts on BFCL, the authoritative global benchmark for large model tool calling, with more than 600,000 downloads on HuggingFace, leading the world. Additionally, she has long served as an Area Chair and reviewer for prestigious conferences and journals such as ICLR, ICML and KDD.

Li Haoran is an Associate Researcher and Master's Supervisor at the Institute of Automation, Chinese Academy of Sciences, as well as a BAAI Young Scholar. His research mainly focuses on deep reinforcement learning and its applications in embodied systems. He has published more than 50 papers in international journals including IEEE TNNLS, TCYB, TSMCS and top conferences such as NeurIPS, ICLR, RSS, CVPR, ICRA and CoRL. He has presided over and participated in the Youth Program, sub-projects of Major Programs, and Key Programs of the National Natural Science Foundation of China. He has won 5 championships and 3 first prizes in multiple domestic and international robotics competitions, and received the Second Prize of Natural Science of the Beijing Science and Technology Award.

Li Ziniu, Ph.D. from The Chinese University of Hong Kong, was supervised by Professor Luo Zhiquan. His research focuses on the efficiency and stability of large-scale reinforcement learning training. He has published more than 20 papers in top machine learning conferences such as ICML and NeurIPS, as well as journals including TPAMI and JASA. His research has received recognitions including the Runner-up Best Paper Award at the NeurIPS FITML Workshop, NeurIPS Spotlight, and ICLR Oral. In addition, his research achievements have been deployed and applied in industrial scenarios at ByteDance, Tencent and other leading tech companies.

Xu Shusheng is a Senior Algorithm Expert at Ant Group and Algorithm Lead of the open-source reinforcement learning framework AReaL. He led the design of AReaL’s fully asynchronous RL training architecture, which achieves a 2.77-fold training acceleration compared with synchronous schemes. Trillion-parameter-level Agent models have been successfully trained based on this framework, reaching state-of-the-art industry standards. The project has gained more than 5,000 GitHub stars since being open-sourced. He obtained his PhD from the Institute for Interdisciplinary Information Sciences and his bachelor’s degree from the Department of Electronic Engineering, Tsinghua University. His research focuses on large language models and reinforcement learning, and he has published over ten papers in top international conferences such as ICML, ICLR, NeurIPS and EMNLP. Among them, his first-author research on RLHF of large models was accepted as an Oral paper at ICML 2024. In this lecture, he will share the system design and practical experience of AReaL in Agentic RL training.

Professor Yue Zhang is currently Vice Dean and Tenured Professor of the School of Engineering, Westlake University. He earned his bachelor’s degree in Computer Science from Tsinghua University, followed by his master’s and doctoral degrees from the University of Oxford, and completed his postdoctoral research at the University of Cambridge. Prior to joining Westlake University in 2018, he worked as an Assistant Professor at Singapore University of Technology and Design (SUTD) from 2012 to 2018.

Li Xiaoxiao is currently an Associate Professor and PhD Supervisor in the Department of Electrical and Computer Engineering at the University of British Columbia (UBC), Director of the UBC Trusted and Efficient AI (TEA) Lab, Canada CIFAR AI Chair, Canada Tier II Research Chair in Trustworthy AI, and Visiting Researcher at Google. Her research interests lie in trustworthy and efficient artificial intelligence, covering the trustworthiness and efficiency of machine learning algorithms and foundation models, AI agent systems, as well as interdisciplinary research between artificial intelligence and cognitive neuroscience. She is committed to developing next-generation trustworthy AI systems to bridge the gap between AI research and practical applications. As first and corresponding author, she has published numerous papers in top-tier international conferences and journals including ICLR, ICML, NeurIPS, CVPR, Nature Methods, IEEE TMI and MedIA. Her representative works include FedBN, BrainGNN, FairMedFM and GMValuator, covering cutting-edge research directions such as trustworthy federated learning, training and evaluation of large language models, data valuation for generative models and time series forecasting. She serves as Area Chair for NeurIPS 2024-2026 and ICLR 2025, and Editorial Board Member of Medical Image Analysis.

Huang Chao is currently Assistant Professor and PhD Supervisor at the University of Hong Kong. His research covers large-scale AI agents, large language models (LLMs) and graph machine learning, with his publications gaining over 17,000 citations on Google Scholar. His team has launched a host of influential open-source projects including nanobot, LightRAG, CLI-Anything, DeepCode, AI-Trader, RAG-Anything, DeepTutor, AutoAgent and AI-Researcher. As the leader of the HKUDS GitHub open-source platform, he has earned more than 270,000 GitHub stars, ranking among the world’s Top 50 and featuring on GitHub Trending over 120 times. In recognition of his contributions to AI open-source development, he was conferred the 2024 WAIC Cloud Sail Bright Star Award and the 2024 ICBS Frontier Science Award, and was selected for the 2025 AI100 Young Pioneers and the 2025 AI 2000 Most Influential Scholars list. His research papers have been repeatedly listed as highly influential papers at top international AI conferences and received numerous Best Paper Nominations.

Tang Jian is currently an Associate Professor at Mila, Quebec AI Institute founded by Turing Award laureate and Godfather of AI Yoshua Bengio, Assistant Professor and doctoral supervisor at the Faculty of Computer Science and Faculty of Business Administration of Université de Montréal, as well as a CIFAR AI Chair. His main research directions include graph representation learning, graph neural networks, knowledge graphs and drug discovery. He won the Best Paper Award at ICML 2014 and received Best Paper Nomination at WWW 2016. He has published a series of classic papers in graph representation learning such as LINE and RotatE. He co-developed TorchDrug, the world’s first open-source machine learning platform for drug discovery with NVIDIA, IBM and Intel. He has repeatedly served as Area Chair for top machine learning conferences ICML and NeurIPS, and Action Editor of the top journal JMLR.

Zheng Shuxin, Vice President of Zhongguancun Institute of Artificial Intelligence, Associate Professor of Beijing Zhongguancun University and Co-Dean of AI Business School, earned his doctoral degree from University of Science and Technology of China. He once served as Principal Researcher at Microsoft Research Asia. His main research focuses on large model algorithms and scientific intelligence. His research findings have been published on the cover of *Science*, subsidiary journals of *Nature* and top international conferences. He has authored more than 30 academic papers with over 7,000 Google Scholar citations and won multiple championships in international AI competitions. In industrial practice, he has promoted the construction of achievement transformation system at Zhongguancun University, incubated more than ten AI technology enterprises with total financing exceeding 300 million yuan and overall valuation surpassing 1 billion yuan. He also holds concurrent positions including external expert member of the Investment Decision Committee of National AI Industry Investment Fund, Adjunct Professor at China University of Political Science and Law, industrial supervisor of Chinese Academy of Sciences and Associate Editor of AI for Science.

Luo Yuyu is an Assistant Professor at The Hong Kong University of Science and Technology (Guangzhou), a Joint Assistant Professor at The Hong Kong University of Science and Technology, and a PhD Supervisor. His research interests focus on autonomous agents and data-centric AI. He has published over 60 papers in top conferences including SIGMOD, VLDB, ICML, ICLR, and KDD. His ongoing projects include a National Major Science and Technology Project and a Young Scientists Fund of the National Natural Science Foundation of China. His honors include the World Artificial Intelligence Conference Yunfan Award, Forbes China 30 Under 30, Best-of-SIGMOD 2023 Papers, Huawei Spark Award, and Outstanding Doctoral Dissertation Awards from Tsinghua University and CCF. He leads the open-source agent projects DeepEye and OpenManus (over 55,000 GitHub stars) and received a Silver Medal at the Geneva Inventions Exhibition. He serves as Workshop Chair for the VLDB 2026 Agentic Data System Workshop and Track Chair for the KDD Cup 2026 Data Agent Track—the first time a Chinese university independently organizes a KDD Cup track. Personal homepage: https://luoyuyu.vip

Chen Siheng is an Associate Professor at the School of Artificial Intelligence, Shanghai Jiao Tong University. He earned his doctorate from Carnegie Mellon University and was selected into the national young talent program, and once worked in the autonomous driving department of UBER ATG. He has undertaken many key research projects including original exploration projects and general projects of the National Natural Science Foundation of China, major AI 2030 projects of the Ministry of Science and Technology and special artificial intelligence projects of Shanghai Science and Technology Commission. He has published over a hundred papers in journals and conferences such as Nature Computational Science, Nature Communications, Cell Patterns, TPAMI, NeurIPS, ICML, ICLR, CVPR, ICCV and KDD, with more than 10,000 Google Scholar citations. He has won many honors including IEEE Signal Processing Society Best Young Author Paper Award, ASME Structural Inspection Society Best Paper Runner-up Award, GlobalSIP 2018 Best Paper Award and Mitsubishi Electric Research Laboratories President’s Award. He developed SciMaster, the world’s first general scientific research agent, which won the Annual Achievement Award of the School of Artificial Intelligence, Shanghai Jiao Tong University. He currently serves as Associate Editor of T-SIPN, Area Chair of NeurIPS, ICML, ICLR and other conferences, and reviewer for Nature Machine Intelligence, and his current research focuses on scientific research agents.

Sun Tianxiang is the Founder and CEO of Analemma, and Assistant Professor at Shanghai Chuangzhi Academy. He has published more than 20 first-author papers in top international conferences including ICML, ICLR, ACL, NAACL and AAAI, with over 5,000 Google Scholar citations. He led the development of MOSS, China's first ChatGPT-like conversational large language model, which has gained 12,000 GitHub stars. FARS, an automated multi-agent scientific research system he developed, secured tens of millions of US dollars in angel round financing from Sequoia China, Gaorong Ventures and other institutions. He has received numerous honors such as the WAIC Cloud Sail Award and ByteDance Scholarship.

Liu Zechun works as Senior Research Scientist and Technical Lead at Meta Reality Labs. Her research focuses on improving the operational efficiency and deployment performance of foundational large models through architecture optimization, low-bit quantization and sparsification techniques, and applies deep learning to address practical industrial technical issues. She has published more than 50 papers in prestigious academic conferences and journals, gaining wide recognition across the academia, with over 9,000 citations on Google Scholar.

Zhang Shaokun is a Research Scientist at NVIDIA Research, focusing on agent training and multi-agent research. He earned his doctoral degree from Pennsylvania State University in 2026, and worked at NVIDIA and Microsoft Redmond Research during his PhD study. His research interests mainly cover agent reinforcement learning and CUA Agent. He has published approximately 20 academic papers, most of which were accepted by three flagship machine learning conferences. Multiple first-author papers were selected for Oral and Spotlight presentations, with over 4,000 citations on Google Scholar. He won the Best Paper Award at the ICLR 2023 Agent Workshop. As the first author of the early self-improving agent AgentOptimizer, his research has been covered by Forbes.

Lin Tao is a Distinguished Researcher, Doctoral Supervisor and Independent PI at the Department of Artificial Intelligence, School of Engineering, Westlake University. He is a recipient of the National Young Talent Program and concurrently serves as a consultant at Ant Technology Research Institute. He received his PhD from École Polytechnique Fédérale de Lausanne (EPFL) in 2022. His research directions include deep learning and optimization, efficient deep learning and inference, as well as distributed deep learning and systems. His paper was nominated for Best Paper Award at ECCV 2024. He has been listed among the World's Top 2% Scientists by Stanford University for two consecutive years from 2024 to 2025.

Gu Yu is Co-founder of NeoCognition, with long-term research on linguistic agents. He earned his PhD in Computer Science from The Ohio State University, and bachelor’s and master’s degrees from Nanjing University. His research papers won Outstanding Paper Awards at ACL 2023 and COLING 2022.

Wang Yan is a Frontier Researcher at Tencent Large Language Model Department, focusing on the research of infinite-long memory model architecture. He previously worked as a research scientist at Tencent AI Lab and miHoYo. As corresponding author, he was awarded the ACL Outstanding Paper Award and received Best Paper Nomination of IEEE Transactions on Games. He developed multiple game agents for long-popular Tencent games including Honor of Kings and Delta Force, supporting tens of thousands of access visits per minute.

Yang Mengyue serves as Assistant Professor at the University of Bristol. She obtained her doctorate from the UCL AI Center under the supervision of Professor Jun Wang. Her research focuses on world models, causality and reinforcement learning. She was honored with KAUST Rising Star in AI 2024 and AAAI 2026 New Faculty Highlights Award. Dr. Yang acts as Area Chair of NeurIPS and Guest Editor for special issues of Journal of Machine Learning. She also works as organizer and Program Chair of embodied world model workshops at ICLR 2024-2025 and NeurIPS 2025.

Wang Liyuan is an Assistant Professor and Doctoral Supervisor at the Department of Psychology and Cognitive Science, Tsinghua University. His research focuses on brain-inspired continual learning theories and methodologies, and promotes their applications in scientific intelligence, smart healthcare and embodied intelligence. His findings have been published as first or corresponding author in top journals and conferences including Nature Machine Intelligence, Nature Communications, TPAMI, Patterns, NeurIPS, ICML and ICLR. He has received honors such as the inaugural Qingyuan Scholar Award from Chinese Association for Artificial Intelligence, World AI Conference Yunfan Award and WAIC Outstanding Youth Paper Award.

Xiong Yuxuan is an Assistant Professor at the Faculty of Artificial Intelligence in Education, Central China Normal University. His research covers AI-enabled education, computer vision, data engineering and computer networks. He won the Best Paper Award at MLMI 2019, and Outstanding Product Awards at the 22nd and 23rd China Hi-Tech Fair.

Min Zhao will join Nanjing University as an assistant professor. She is a core developer of the domestic video generation model Vidu and a recipient of Tsinghua University's Shuimu Scholar Program. Her research focuses on video generation and interactive video world models. She has published numerous papers as first author and co-first author in top international conferences including NeurIPS, ICML, and ICLR. Her first-authored papers have garnered over 700 citations, with total citations exceeding 1,000. Her proposed video length extrapolation method RIFLEx and real-time interactive generation method Casual Forcing have been integrated into mainstream open-source video diffusion frameworks such as HunyuanVideo and CogVideoX. Open-source projects led by her have accumulated nearly 3,000 GitHub Stars.

Mingyang Deng, a PhD candidate at MIT, studies under Kaiming He, creator of ResNet. His research focuses on generative models. He is the first Chinese national to claim gold medals in both IMO and IOI, and the third perfect scorer in IOI history. He earned his bachelor's degree in Mathematics and Computer Science at MIT, and completed research internships at DeepMind and Meta. In 2026, he proposed Drifting Models as the first author. This innovative image generation approach produces high-definition images in a single step, enables fast rendering on mobile devices and lowers barriers to AI creation.

Ailing Zeng, Researcher at Anuttacon, leads the development of LPM 1.0, a real-time conversational video generation system for interactive character performance. Her primary research focuses on a full-stack technical architecture spanning human-centric perception and understanding to long-horizon controllable generation. She has made deep contributions to long-term time series forecasting, with her representative work LTSF-Linear receiving the AAAI 2023 Most Influential Paper Award, establishing foundational theoretical contributions to temporal modeling and long-period prediction. She has published over 50 papers in top international academic conferences and spearheaded multiple open-source projects, accumulating more than 25,000 GitHub Stars and over 13,000 Google Scholar citations. She earned her Ph.D. in Computer Science and Engineering from The Chinese University of Hong Kong and served as a visiting scholar at the Robotics Institute of Carnegie Mellon University. She is committed to building video-native multimodal intelligent systems, aiming to enable human cognition, interaction, and content generation with the physical world, and to create intelligent interactive agents capable of autonomous evolution and logical self-consistency in open-ended interactive environments.

Sida Peng, Researcher under the "Hundred Talents Program" at the School of Software Technology, Zhejiang University, and Ph.D. Advisor. His research focuses on 3D computer vision and computer graphics. To date, he has published 9 papers in TPAMI/TOG and 15 Oral/Highlight papers at CCF-A ranked conferences. He has garnered over 9,000 Google Scholar citations, with one first-authored paper receiving a CVPR Best Paper Nomination. His work has accumulated tens of thousands of GitHub Stars and earned the 2024 CCF China Excellent Graphics Open Source Software Award. He has been selected as a China3DV 2025 Outstanding Young Scholar, named to Stanford's 2024/2025 World's Top 2% Scientists list, and received the 2024 CCF China Outstanding Doctoral Dissertation Award (one of ten selected nationwide in computer science). He was named a 2022 Apple Scholar (the only recipient in the Asia-Pacific region).

Tengfei Wang received his Ph.D. from the Hong Kong University of Science and Technology. He currently leads the development and deployment of Hunyuan World Models at Tencent, having built the data and algorithmic R&D infrastructure for Hunyuan World Models from the ground up. Under his leadership, the team has successively released multiple models including HY World 1.0, 2.0, WorldPlay, and WorldMirror, garnering widespread attention from the open-source community. He has published over 40 papers in top-tier AI journals and conferences, with more than 3,500 Google Scholar citations. His series of open-source projects have accumulated over 20,000 GitHub Stars, and his research has been recognized with Most Influential Paper awards at ICCV and ECCV.

Yu Cheng, Chief Scientist at Kunlun Tech and Associate Professor at the Department of Computer Science and Engineering, The Chinese University of Hong Kong. From 2018 to 2023, he served as a Principal Researcher at Microsoft Research Redmond. Prior to joining Microsoft, he was a researcher at IBM Research and the MIT-IBM Watson AI Lab. His research spans deep learning, with a particular focus on model compression and efficiency, deep generative models, and language/multimodal large models. Starting in 2021, he led his team to work closely with OpenAI on efficiency, robustness, and scalability optimizations for the GPT series of models, driving the productization of related services and applications, including New Bing powered primarily by GPT-4, GitHub Copilot backed by GPT-3.5, and Image Creator supported by DALL-E 2. From 2023 to 2025, he led or contributed to the development of products and models including Minimax abab6.5, the M1/Hailuo Video model, as well as Skywork R1V2/V3 and Super Agent.