Bio

I am currently an ML research at Raidium where I work on adapting large-scale vision–language and self-supervised foundation models to the medical domain. My current research focuses on bridging the gap between natural and medical imaging by extending self-supervised learning to CT and MRI data, with the goal of learning transferable, robust representations for clinical tasks. I am also working also contrastive reports to 3D MRI/CT volumes pre-training as a backbone for report generation.

Previously, I interned at Valeo.ai as part of my MVA Master’s degree at ENS Paris-Saclay in France. I studied multi-camera tokenisation methods applied to the field of autonomous cars. I am genuinely motivated by research that bridges theory and practice, bringing theoretical advances into real-world applications.

News

• News 📄 (08/09/2025): Curia new family of SOTA VIT radiological foundation model for MRI/CT images is out arxiv.

• News 🩻 (10/03/2025): Joined Raidium as a permanent ML Researcher.

• News 🎓 (01/09/2024): Obtained my MS. MVA from ENS Paris-Saclay with the highest honor (18.20/20.0).

• News 🚗 (15/04/2024): Joined Valeo.ai research team for a 6-month intership til October on “Tokenization from multi-camera autonomous drivin data” under the supervision of Florent Bartoccioni and Spyros Gidaris.

GitHub

@b-ptiste — Open-source projects, ML/CV code, and data challenge repos.

Data Challenge Achievements

• Text-to-Molecule (Polytechnique) – 🏆 1st out of 52
• Video Retrieval (University of Klagenfurt) – 🏆 1st out of 5
• Few-Shot Novelty Instance Segmentation (Collège de France) – 🥈 2nd out of 20
• Sketch Classification (ENS Ulm) – 🏅 4th out of 59
• Multiple Instance Learning (CentraleSupélec - Université Paris-Saclay) – 🏅 16th out of 39

Projects