Tom Röhr

Tom Röhr is a PhD researcher at the Data Science and Text-based Information Systems (DATEXIS) group at the Berliner Hochschule für Technik (BHT) and a member of the Data Science +X Graduate School. His research focuses on natural language processing and large language models, particularly reasoning, multilinguality, post-training, and applications in healthcare.

He is part of SOOFI (Sovereign Open Source Foundation Models), where he is in a leading position for the post-training efforts. His work focuses on developing and improving open foundation and reasoning models through methods including reinforcement learning, reasoning-oriented post-training, multilingual adaptation, and evaluation.

His recent research covers multilingual reasoning in large language models, reinforcement learning with verifiable rewards, medical reasoning, clinical NLP, and the analysis of doctor–patient interactions.