[Jobs] PhD/Postdoc Positions on LLM Architectures, Interpretability, and Safety at Saarland University

[Michael H]

The Language, Computation, and Cognition Lab (LaCoCo) at Saarland Informatics Campus (Saarbrücken, Germany), directed by Michael Hahn, invites applications for fully funded PhD and postdoc positions, with a flexible start date.


RESEARCH AREAS

We’re especially excited about projects on:

• Architectural limitations of large language models (LLMs), especially from theoretical perspectives
• New architectures and efficient reasoning for LLMs
• Mechanistic interpretability
• Theoretical foundations for AI safety

You’ll have substantial freedom to shape your topic within the lab’s scope.


ABOUT THE LAB

The lab is generously supported by an Emmy Noether grant (DFG) and has substantial GPU resources. We regularly publish at top venues (e.g., six NeurIPS/ICML/ICLR papers in 2024–2025; ACL Best Paper 2024) and are embedded in a world-class environment: Saarland University (Departments of Computer Science and Language Science & Technology), MPI for Informatics, MPI-SWS, CISPA, and DFKI.

Learn more about the lab at https://lacoco-lab.github.io/

HOW TO APPLY

Please follow the procedure here: https://lacoco-lab.github.io/joining/

Deadline: December 20, 2025.

For questions, please contact Michael Hahn, mh...@lst.uni-saarland.de

Applicants are expected to have a Master's degree by the start date. Applicants who only have an undergraduate degree should instead apply to the Saarbrücken Graduate School of Computer Science.

-- 

Michael Hahn

Assistant Professor

Saarland Informatics Campus

Saarland University

Group: https://lacoco-lab.github.io/

Personal: https://www.mhahn.info/