I work on LLMs evaluation, tasks and dataset design and development,and interpretability, with a focus on representational engineering of LLMs. I’m also interested in the intersection of LLMs and knowledge graphs, and how to leverage structured knowledge to improve LLMs performance and interpretability.