I am an ML researcher at Microsoft Research Cambridge working on generative models. Before, I did my PhD in Statistical Machine Learning in the OxCSML group at the University of Oxford, supervised by Prof. Chris Holmes and Prof. Arnaud Doucet. I was also a Graduate Teaching and Research Scholar in Computer Science at Oriel College, University of Oxford, teaching maths courses, and studied at The Alan Turing Institute, the UK’s national institute for AI.

My research focusses on:

• Understanding the inductive biases of generative models, particularly diffusion/flow-matching models, hierarchical VAEs, and their neural architecture such as the U-Net
• Properties of LLMs, such as their fixed-point behaviour, counterfactual reasoning, and their ability to perform Bayesian inference
• Large-scale, multi-modal applications of generative models, such as analog-amenable machine learning, inverse problems, and radiology-report generation

Research highlights include a NeurIPS 2022 oral (top 1.76%), an ICML 2025 spotlight (top 2.6%), 2 workshop orals at ICLR and AAAI, a publication in Nature Reviews Genetics and a Best Paper Award at TAROS, the UK’s largest robotics conference before my PhD.

If you would like to collaborate or be co-supervised by me, please reach out!

Contact: fabian.falck1@gmail.com