About Me
I am an Assistant Professor at the Warwick Manufacturing Group (WMG), University of Warwick. My research focuses on safe and probabilistic generative models, including robustness, uncertainty, and alignment in large-scale AI systems. Previously, I was a postdoctoral researcher at the University of Manchester, working with Prof. Samuel Kaski and supported by the UKRI Generative AI Hub. I completed my PhD at the University of Cambridge, supervised by Dr. Damon Wischik and supported through a personal grant from Mrs. Yanshu Wu. My educational background includes a BSc in Computer Science and Mathematics from Manchester, followed by an MPhil in Computer Science and an MASt in Mathematics from Cambridge. For a complete overview of my academic and professional journey, please refer to my CV.
I'm passionate about learning and creative problem-solving. I'm always open to collaboration - feel free to reach out to me at az381 [at] cantab [dot] ac [dot] uk.
I am currently recruiting PhD students for Fall 2027 entry, and I also welcome research interns. Prospective students and interns interested in probabilistic modelling, safe large language models, and AI safety are encouraged to get in touch. If you are interested in applying for a PhD, please fill in this questionnaire. If you are interested in a research internship, please email me your CV, transcript, and a brief description of your research interests.
Selected Publications
Concept-based Adversarial Attack: a Probabilistic Perspective
[paper] [code] [poster]
Andi Zhang, Xuan Ding, Steven Mcdonagh, Samuel Kaski.
ICLR 2026
Your Finetuned Large Language Model is Already a Powerful Out-of-distribution Detector
[paper] [code] [poster] [slides]
Andi Zhang, Tim Z. Xiao, Weiyang Liu, Robert Bamler, Damon Wischik.
AISTATS 2025 & UKAIRS 2025 Oral
Constructing Semantics-Aware Adversarial Examples with a Probabilistic Perspective
[paper] [code] [poster]
Andi Zhang, Mingtian Zhang, Damon Wischik.
NeurIPS 2024
Falsehoods that ML researchers believe about OOD detection
[paper] [poster]
Andi Zhang, Damon Wischik.
NeurIPS 2022 Workshop
On the Out-of-Distribution Generalization of Probabilistic Image Modelling
[paper] [code]
Mingtian Zhang*, Andi Zhang*, Steven McDonagh.
NeurIPS 2021