As an AI researcher with a background in mathematics and physics, I specialize in training large language models (LLMs) from scratch. My expertise covers the entire LLM pipeline, from data preprocessing to model deployment. I'm a core contributor to the open-source library Giotto-deep, which integrates topological tools to study and enhance deep learning models. As a mentor at DeepLearning.AI and a speaker at AI community meetups, I guide learners, share knowledge, and contribute to the collective advancement of the field. I also deliver topological data analysis and machine learning talks at institutions such as Inria Paris, the University of Oxford, and the Technical University of Munich.