I am a postdoc at Mila (Quebec AI institute) and McGill University, with Prof Siva Reddy. I obtained my PhD from the UK Research & Innovation Centre for Doctoral Training in Natural Language Processing at the University of Edinburgh, supervised by Prof Dr Ivan Titov. Before that, I studied at the University of Amsterdam for my MSc and BSc in Artificial Intelligence, and I was an intern at Meta AI Research (2022) and Microsoft Research (2024).
My primary research focus currently lies on understanding the memorization vs generalization balance in LLMs, by analyzing them using data from the outside, or via interpretability techniques on the inside. Throughout my PhD, I primarily focused on this through the lens of the (non-)compositionality of language. Natural language is largely compositional, but also contains figurative and formulaic subsets, for which humans and machines both rely more heavily on memorization. Other topics I'm always happy to talk about are interpretability more generally, multilinguality, tokenization and (psycho)linguistics-inspired AI.