Jun-Yan Zhu

Jun-Yan Zhu is the Michael B. Donohue Assistant Professor of Computer Science and Robotics at Carnegie Mellon University's School of Computer Science. He holds affiliated faculty roles in the Computer Science Department and Machine Learning Department. His research focuses on generative models, computer vision, graphics, and computational photography.

Education:

• Ph.D., UC Berkeley (2017), advised by Alexei A. Efros
• B.E., Tsinghua University (2012), advised by Zhuowen Tu, Shi-Min Hu, and Eric Chang
• Postdoc, MIT CSAIL (2017–2018), with William T. Freeman, Josh Tenenbaum, and Antonio Torralba

Research Interests: Zhu's work explores the synergy between human creators and generative models. Key areas include:
- Controllable Visual Synthesis: Developing algorithms for precise image/video editing and generation.
- Model Customization: Enabling users to adapt models for new tasks/concepts with minimal input.
- Data Attribution: Addressing ethical challenges in synthetic data usage.
- 3D/Neural Rendering: Advancing techniques for 3D object synthesis and tactile integration.
His lab, Generative Intelligence Lab, emphasizes human-AI collaboration and practical applications like NVIDIA Canvas and Adobe Firefly.

Articles Trends: Recent work spans 3D object generation (LEGO designs), efficient diffusion models (SVDQuant), tactile-aware 3D synthesis, and ethical data attribution systems. His research balances technical innovation with user-centric design, often bridging theory and industry applications.

Awards:

• ACM SIGGRAPH Outstanding Doctoral Dissertation (2017)
• CVPR Best Paper Finalist (2022), ICRA Best Paper (2024)
• NVIDIA GTC Best in Show (2019) for GauGAN

Advising & Grants: Supervises 10+ PhD students across CMU Robotics (RI), Machine Learning (MLD), and Computer Science (CSD). Active in NSF grants and industry collaborations (Adobe, NVIDIA). His lab hosts the Generative Intelligence Lab, part of CMU Graphics Lab and Computer Vision Group.