Hongyang Zhang

Hongyang Zhang is an Assistant Professor at the University of Waterloo's David R. Cheriton School of Computer Science (part of the Faculty of Mathematics) and a faculty member of Vector Institute for AI. His research focuses on machine learning theory and applications, including inference acceleration for large language models (e.g., EAGLE series), world models for robotics and autonomous systems, and AI security. He leads the SafeAI Lab and is affiliated with the AI Institute and Cybersecurity and Privacy Institute.

Education:

• Ph.D. in Machine Learning, Carnegie Mellon University (2019)
• Postdoc at Toyota Technological Institute at Chicago (2019–2021)
• Bachelor's degree from Peking University (2015)

Research Interests:

• Inference acceleration (e.g., EAGLE-3 achieving 5× speedup)
• World models for infinite-horizon video generation (The Matrix)
• AI security, adversarial robustness, and watermarking
• System-2 LLMs for alignment and reasoning

Awards & Recognition:

• 1st place in multiple adversarial vision challenges (NeurIPS 2018, CVPR 2021)
• AAAI New Faculty Highlights (2023)
• IEEE Senior Member (2024)
• Amazon Research Award and WAIC Yunfan Award

Academic Leadership:

• Area Chair for ICML, NeurIPS, ACL, and ICLR
• Action Editor for Data-centric Machine Learning Research (DMLR)
• Teaching: Introduction to ML, Robustness of ML, and AI Security courses

Labs & Teams: Leads the SafeAI Lab, collaborating on projects like EAGLE, The Matrix, and zkLLM.