VLM Agents · Post-training · Embodied AI

Zengyi Wo

Joint Ph.D. Student, Nanjing University & CASIA · PRLab

I study how post-training helps vision-language agents reason, use tools, and act in the physical world.

Zengyi Wo standing in a garden pavilion

Biography

I am a joint Ph.D. student at Nanjing University and the Institute of Automation, Chinese Academy of Sciences (CASIA), affiliated with the Pattern Recognition Laboratory (PRLab). I am advised by Prof. Chenyang Si and Prof. Liang Wang.

My research connects VLM agents and post-training with Embodied AI: improving multimodal foundation models for reasoning, tool use, and planning, then grounding these capabilities in embodied systems for manipulation, navigation, and long-horizon decision making.

Before my Ph.D., I worked as an Algorithm Engineer at Baidu on multimodal large language model post-training, including SFT, RLHF, RLVR, and OPD for mathematical and visual reasoning. I received my M.S. in Computer Science from Tianjin University in 2025 and my B.S. from Zhengzhou University in 2022. I was also a Research Intern at Duxiaoman Finance in 2025, working on applied intelligent systems.

Research Interests

VLM Agents Vision-language agents for multimodal perception, tool use, planning, reasoning, and grounded decision making.
Post-training SFT, RLHF, RLVR, alignment, reward modeling, and multimodal reasoning for stronger foundation-model agents.
Embodied AI Physical interaction, manipulation, navigation, embodied reasoning, and autonomous agents that operate in dynamic environments.
Seeking AI Lab / Big Tech research internships I am actively seeking research internships at leading AI labs and top technology companies, especially around VLM agents, post-training, and Embodied AI. I am also open to research collaborations. If my work aligns with your interests, please feel free to reach out.
Email me

Advisors & Collaborators

I am affiliated with PRLab at Nanjing University and jointly trained with the Institute of Automation, Chinese Academy of Sciences (CASIA). I am advised by Prof. Chenyang Si and Prof. Liang Wang. I am also grateful to collaborate with outstanding researchers, including Prof. Lu Lin at Penn State and Prof. Ruijie Wang at UIUC.

Student Collaborators

I am fortunate to work with motivated student collaborators on topics around VLM agents, post-training, graph learning, and Embodied AI. I especially enjoy hands-on collaboration on research ideas, experiments, paper writing, and open-source projects.

VLM Agents Shiyu Zhang
Post-training Xin Deng · Jingchen Hao · Xuexin Liu · Siyao Qiao
Graph Learning Yuhang Liu · Yunlong Chu · Bing Hao
Interested in working together? Students who are interested in related directions are welcome to reach out by email.

News

Selected Publications

  1. SIGIR’26
    Learning Noise-Resilient and Transferable Graph-Text Alignment via Dynamic Quality Assessment
    Yuhang Liu, Minglai Shao, Zengyi Wo, and 5 more authors
    In Proceedings of the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval, 2026
  2. AAAI’26
    LLMTM: Benchmarking and Optimizing LLMs for Temporal Motif Analysis in Dynamic Graphs
    Bing Hao, Minglai Shao, Zengyi Wo, and 3 more authors
    In Proceedings of the 40th AAAI Conference on Artificial Intelligence, 2026
  3. IJCAI’25
    Leveraging Personalized PageRank and Higher-Order Topological Structures for Heterophily Mitigation in Graph Neural Networks
    Yumeng* Wang, Zengyi* Wo, Wenjun Wang, and 2 more authors
    In Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence, 2025
    *Equal contribution
  4. KDD’25
    Local Homophily-Aware Graph Neural Network with Adaptive Polynomial Filters for Scalable Graph Anomaly Detection
    Zengyi Wo, Minglai Shao, Shiyu Zhang, and 1 more author
    In Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2025
  5. DASFAA’25
    Mining Denoising Complementarity and Consistent Consensus for Unsupervised Multiplex Graph Representation Learning
    Chang Liu, Zengyi Wo, Minglai Shao, and 2 more authors
    In Database Systems for Advanced Applications, 2025
  6. WWW’24
    Graph Contrastive Learning via Interventional View Generation
    Zengyi Wo, Minglai Shao, Wenjun Wang, and 2 more authors
    In Proceedings of the ACM Web Conference 2024, 2024

Honors & Awards

  • AAAI 2026 Oral Presentation
  • KDD 2025 Oral Presentation
  • WWW 2024 Oral Presentation
  • National Scholarship for Graduate Students, 2021
  • Outstanding Graduate Student Award, Tianjin University, 2025

Academic Services

  • Reviewer: KDD, AAAI, IJCAI, TKDD
  • Program Committee Member: WWW