Physics-Constrained Fine-Tuning of Flow-Matching Models for Generation and Inverse Problems
ICLR 2026
PhD Researcher in Scientific Machine Learning
Hi, I’m Jan Tauberschmidt.
I am excited about machine learning methods that become useful in scientific and physical systems. This is the direction I explore as a Ph.D. student at RPTU Kaiserslautern and researcher at the German Research Center for Artificial Intelligence (DFKI), working in the Department of Data Science and its Applications under the supervision of Prof. Sebastian J. Vollmer.
Research direction
My research asks how expressive machine learning models can be made useful for scientific problems without ignoring the structure we already know. I work on generative modeling, physics- and domain-informed learning, and control, aiming to bridge the gap between black-box prediction and classical mathematical modeling.
A recent piece of this puzzle
We may have crossed paths in…
2026
Rio de Janeiro
ICLR 2026
2025
London
Visited Imperial College and Alan Turing Institute
2022, 2024
Porsche Motorsport, Weissach
Internship and Master’s thesis
2018-now
Kaiserslautern
DFKI (Researcher) Fraunhofer ITWM (Research Assistant) RPTU (Bachelor's and Master's of Mathematics)