Before founding Adaptive ML, Julien led the extreme-scale research teams at HuggingFace and LightOn, where he played a pivotal role in developing state-of-the-art open-source models. His contributions pioneered advances in data scalability & quality, large-scale training pipelines, and principled model architectures. He is also the organizer of the Efficient Systems for Foundation Models workshop at ICML, a forum fostering discussions on scalable AI systems.