Gaspard Beaudouin

ENS Paris-Saclay. École nationale des ponts et chaussées, Institut Polytechnique de Paris.

pic-gaspard.jpg

I am currently a Research Scientist Intern at Mistral AI on the Multimodal team, working on Vision Encoder and multimodal models. Upon completion, I will earn a Master’s degree in Mathematics, Vision and Learning (MVA) from ENS Paris-Saclay and graduate from École nationale des ponts et chaussées, where I specialized in Mathematics and Computer Science.

I have fairly broad research interests, among them VLM pre-training, Representation Learning (JEPA, SSL for videos, …), World Models, flow matching and reasoning. I conducted research on flow matching at Harvard AI and Robotics Lab, supervised by Prof. Mengyu Wang. I have also worked on Large Language Models during an internship at ChapsVision, more precisely about training for agentic/tool calling abilities.

I am the first author of Delta Rectified Flow Sampling for Text-to-Image Editing, accepted at CVPR 2026. I also co-authored SplitFlow: Flow Decomposition for Inversion-Free Text-to-Image Editing, accepted at NeurIPS 2025. Thanks to Minghan Li, Sung-Hoon Yoon, Jaeyon Kim and Mengyu Wang for their incredible collaboration. I also co-authored PAGE-4D: Disentangled Pose and Geometry Estimation for 4D Perception, accepted at ICLR 2026, with Kaichen Zhou. I also contributed to Mistral Large 4.

You can find my code and projects on GitHub.

news

🚀 Mistral Large 4 is out! Mistral AI announced Mistral Large 4 (ML4 / “le Chonk”), a natively multimodal open-weight model. The technical report is now out.
🐈 I joined Mistral AI as a Research Scientist Intern on the Multimodal team, working on Vision Encoder.
🚨 News: Our paper “Delta Rectified Flow Sampling for Text-to-Image Editing” has been accepted at CVPR 2026. Read the paper
🧠 PAGE-4D accepted at ICLR 2026! Our paper “PAGE-4D: Disentangled Pose and Geometry Estimation for 4D Perception” has been accepted at ICLR 2026. Read the paper
🎓 Starting Master’s Program! I’m excited to begin the Master MVA program at ENS Paris-Saclay in fall 2025, focusing on Computer Vision, Machine Learning and Natural Language Processing.
🎉 SplitFlow accepted at NeurIPS 2025! Our paper “SplitFlow: Flow Decomposition for Inversion-Free Text-to-Image Editing” has been accepted at NeurIPS 2025! Read the paper
🔬 Research Internship at Harvard! I had the opportunity to conduct research on flow matching and generative models at Harvard University, working on novel approaches to flow-based generative modeling.