Dr Gaojie (Jay) Jin (Trustworthy AI)

I am a Lecturer (Assistant Professor) in Computer Science at the University of Exeter, previously I worked at the Chinese Academy of Sciences and the University of Liverpool. I earned my PhD from the University of Liverpool.

I am interested in developing reliable (e.g., robust, generalisable, and efficient [NeurIPS20,CVPR22,23,ICCV23a,TIFS25,ICLR25a]) algorithms for modern machine learning models and applications [ICCV23b,AAAI23,ECCV24]. I focuse on aligning AI with human rules/preferences [ICML24,ICLR25b], with an emphasis on providing theoretical guarantees and statistical analyses [TMLR22,TPAMI25].

Currently, my research interests include, but are not limited to:

1. Trustworthy Human–GenAI Alignment: Developing unified statistical frameworks, e.g., combining PAC-Bayesian theory, conformal prediction, and adversarial analysis, to provide rigorous guarantees for human–GenAI alignment.
2. Efficiency Robustness of GenAI: Building empirical and theoretical foundations to analyse and mitigate vulnerabilities in reasoning efficiency.

I am pleased to announce multiple openings for PhD and internship positions for the 2025/26 academic year. Scholarships are available through the University of Exeter and the CSC-Exeter program. If you are interested in these opportunities or would like to discuss potential collaborations, please don’t hesitate to contact me at g.jin@exeter.ac.uk.

Please find my personal page: https://alexkael.github.io/