Research Fellow in Few step Generative Modelling

University College London (UCL) is a top-ranked computer science research institution collaborating with the University of Cambridge through the AI Hub in Generative Models. The Research Fellow will investigate computationally efficient few-step generative models, develop stable training and inference methods, conduct probabilistic modelling research, perform large-scale computational experiments, publish papers, and co-supervise postgraduate students.

Responsibilities:

• Successful applicants will investigate few-step generative models, including diffusion, consistency, flow-based, and related probabilistic methods, with the aim of developing computationally efficient and theoretically grounded approaches that maintain high-quality generation performance while reducing sampling cost
• The work will involve developing stable training and inference methods, conducting probabilistic modelling research, and performing large-scale computational experiments using standard machine learning frameworks and publicly available scientific datasets
• Key duties and responsibilities include working on deep learning, probabilistic modelling, transformer-based models, diffusion models, flow matching or consistency models, publishing papers and co-supervising PhD and master students