Mission. I am interested in the training and adaptation of large language models across the full lifecycle, including pre-training, post-training, and downstream specialization, with a particular focus on reasoning behavior. I study how reasoning capabilities emerge through large-scale pre-training, and how correctness, verification, and robustness are enforced through post-training and alignment procedures.

Moreover, I actively research the mechanistic interpretability of large language models, for example by tracing information flow and analyzing internal model components, with the goal of understanding and controlling reasoning-related representations and decision processes.

A central theme of my work is the development of reasoning-oriented language models aligned with European AI sovereignty.

Timeline.

2023 - now:	Senior Researcher at Hessian.AI, Darmstadt, Germany.
2021 - now:	Postdoc at the Machine Learning Lab, CS Department, TU Darmstadt, Germany
2017 - 2021:	Ph.D. Student at the UKP Lab and Machine Learning Lab, CS Department, TU Darmstadt, Germany