With a background in theoretical computer science, I work at the intersection of formal methods and artificial intelligence. I apply formal methods to ensure trustworthiness and explainability in AI systems and machine-learning algorithms, particularly sequence models, including language models and time-series predictors. Part of this research is carried out in the PEPR IA SAIF project and our FM4AI group at LMF, where we develop rigorous methods for verifying and improving AI reliability.

I recently gave a talk on explainability, verifiability, and the AI Act, with a focus on what formal methods can contribute. You can find the pdf file here.