PhD in Life Science

The research will be conducted within the MIST project (Scalable Mechanistic Interpretability for Safe and Trustworthy LLM Agents). The project aims to develop scalable methods for understanding the inner workings of large language models (LLMs) and LLM agents, focusing on identifying causal mechanisms underlying tool use, reasoning, and multi-agent communication. The research will investigate cross-model and cross-lingual universality of these mechanisms and develop functionally grounded steering techniques and methods to issue safety certificates.