Hao Wu
Portrait of Hao Wu

Hao Wu

Researcher @ Tencent, @ Department of Computer Science, University of Science and Technology of China

Email GitHub Google Scholar · 1000+ Citations

I am currently a Researcher at Tencent. I graduated from the Department of Computer Science at the University of Science and Technology of China (USTC). During my master's studies, I was also a joint training student in the large model training group of the Machine Learning Platform Department at Tencent.

My research lies at the intersection of scientific AI, multimodal large language models, agent systems, robot video / world models, and large-scale spatiotemporal forecasting. More broadly, I am interested in building intelligent systems that can understand, predict, and reason about the physical world across space, time, and modalities.

My recent and future focus includes four closely related directions: agentic reasoning, multimodal large models, robot video generation and embodied world models, and large-scale spatiotemporal forecasting. My work has appeared in venues such as ICLR, NeurIPS, ICML, KDD, AAAI, ICCV, ACM MM, TKDE, and TPAMI, with nearly 30 CCF-A publications.

Feel free to reach out via email if you are interested in discussing research ideas or potential collaborations.

Current Research Interests

Robot video and world models
Robot World Models Studying robot video generation, reward-aligned training, and long-horizon embodied prediction.
Agent systems
Agent Systems Building agentic systems for scientific and real-world reasoning with retrieval, planning, and tool use.
Multimodal large language models
Multimodal LLMs Exploring multimodal understanding, reasoning, alignment, and safety-aware learning.
Large-scale spatiotemporal forecasting
Spatiotemporal Forecasting Advancing large-scale forecasting for weather, ocean, and Earth systems with physically grounded models.

News

2026.01
Two papers were accepted to TPAMI. Congrats to all collaborators!
2025.12
Two papers were accepted to AAAI and ICLR. Congrats to all collaborators!
2025.06
As corresponding author, one paper was accepted to ICCV. Congrats to all collaborators!
2025.06
Graduated from the joint training program between USTC and Tencent, and received the Outstanding Graduate honor.
2025.05
As co-first author, one paper was accepted to ICML. Congrats to all collaborators!
2025.03
As first author, one paper was accepted to KDD. Congrats to all collaborators!
2024.12
As corresponding author, one paper was accepted to ICLR. Congrats to all collaborators!
2024.07
As first author or co-first author, three papers were accepted to NeurIPS. Congrats to all collaborators!
2024.05
As first author, one paper was accepted to KDD. Congrats to all collaborators!
2024.05
As first author, one paper was accepted to ACM MM. Congrats to all collaborators!
2024.05
As first author, one paper was accepted to ICML. Congrats to all collaborators!
2023.12
As first author, two papers were accepted to AAAI. Congrats to all collaborators!
2023.07
As co-first author, one paper was accepted to NeurIPS 2024. Congrats to all collaborators!
2022.10
Received the First-Class Scholarship at the University of Science and Technology of China.
2022.09
Received the National Scholarship at the University of Science and Technology of China, ranking in the top 1% in China.

Experience

Researcher, Tencent Tianyan Lab

Aug. 2025 - Present

Tencent Tianyan Lab

Continuing research on multimodal foundation models, agent systems, and scientific intelligence in industrial-scale settings.

Research Intern, Tencent Hunyuan

Aug. 2023 - Jul. 2025

Machine Learning Platform Department

Worked on large models, scientific machine learning, world models, and multimodal generative modeling in Tencent Hunyuan.

Online Research Intern, UCLA

May 2023 - May 2024

Advisor: Xiao Luo

Conducted remote research on multimodal learning, dynamics modeling, and related machine learning problems.

Research Intern, HKUST (Guangzhou)

Mar. 2023 - Aug. 2023

Advisors: Yuxuan Liang and Kun Wang

Worked on spatiotemporal modeling and early-stage scientific AI research in the CityMind research environment.

Selected Publications

Full list on Google Scholar
Robot world model result gif

Distilled Multimodal Rewards for Reinforcement Learning of Robot Video World Models

Hao Wu et al. First Author
Ongoing
Safety-aware rollouts paper figure

Safety-Aware Rollouts with Self-Reflection and Structured Rewards

Huahui Yi, Kun Wang, Haolong Hu, Moayad Aloqaily, Liang Lin, Junhao Dong, Qiankun Li, Xing Fan, Hao Wu, Yang Liu, and Qingsong Wen Corresponding Author
TPAMI, under review
TritonCast paper figure

TritonCast: Advanced Long-term Earth System Forecasting

Hao Wu, Yuan Gao, Ruijian Gou, Xian Wu, Chuhan Wu, Huahui Yi, Johannes Brandstetter, Fan Xu, Niklas Boers, Kun Wang, Penghao Zhao, Hao Jia, Qi Song, Xinliang Liu, Juncai He, Shuhao Cao, Huanshuo Dong, Yanfei Xiang, Fan Zhang, Haixin Wang, Xingjian Shi, Qiufeng Wang, Shuaipeng Li, Ruobing Xie, Feng Tao, Yuxu Lu, Yu Guo, Yuntian Chen, Yuxuan Liang, Qingsong Wen, Wanli Ouyang, Deliang Chen, and Xiaomeng Huang
NMI, under revisions
ICCV paper figure

Frequency-Aligned Knowledge Distillation for Lightweight Spatiotemporal Forecasting

Yuqi Li, Chuanguang Yang, Hansheng Zeng, Zeyu Dong, Zhulin An, Yongjun Xu, Yingli Tian, and Hao Wu Corresponding Author
ICCV 2025

Service