Hang Liu

Hang Liu

Third-year undergrad @ XMU

Home

Hi there, I’m Hang Liu! 👋

About Me

I am a third-year (Class of 2023) undergraduate student at the Department of Artificial Intelligence, School of Informatics, Xiamen University.

📬 Email: liuhang1@stu.xmu.edu.cn

I am fortunate to work as a research intern at the MAC Lab (Key Laboratory of Multimedia Trusted Perception and Efficient Computing), Xiamen University, under the supervision of Dr. Jiayi Ji and Prof. Rongrong Ji. Previously, I worked as a research assistant with Prof. Di Hu at the Gaoling School of Artificial Intelligence, Renmin University of China.

My research focuses on multimodal learning and generative models, with a particular interest in exploring the fundamental relationship between understanding and generation. I am working towards developing unified models that bridge multimodal understanding and generation, investigating how these two capabilities can mutually enhance each other to achieve more general and robust multimodal intelligence.

I am actively looking for Ph.D. or Master’s opportunities starting Fall 2027.

💡 I believe clarity is the key to both research and coding.

📰 News

  • [Oct 2025] 🏆 Awarded National Scholarship!
  • [Jun 2025] 🎉 Our paper RollingQ has been accepted to ICML 2025!
  • [Jun-Sep 2025] 💼 Interned at DeepRoute.ai, Shenzhen, focusing on model R&D and optimization for autonomous driving
  • [Mar 2025] 🔬 Started research internship at MAC Lab, Xiamen University, supervised by Dr. Jiayi Ji and Prof. Rongrong Ji
  • [Sep 2024] 📚 Joined as Research Assistant at Gaoling School of AI, Renmin University of China, working with Prof. Di Hu

Research Interests

My research interests lie at the intersection of multimodal perception and generation:

  • Multimodal Large Language Models (MLLMs): Exploring efficient architectures and training strategies for cross-modal understanding and reasoning
  • Generative Models: Developing unified frameworks for multimodal content generation, including diffusion models and autoregressive approaches
  • Unified Multimodal Understanding & Generation: Building systems that can seamlessly perceive, understand, and generate content across vision, language, and audio modalities

Contact

Feel free to reach out for research collaboration or discussion!

Thanks for visiting! I’m always open to exciting research opportunities and collaborations. 😊