Ciao 👋
I’m an AI Scientist at Mistral in Paris. My research focuses on multimodal foundation models for vision and language. Previously, I was a postdoctoral researcher at FAIR in Meta AI, where I contributed to the development of DINOv3, a state-of-the-art self-supervised vision foundation model, and DINO-world, a latent video world model. During my PhD at KTH, Stockholm, I worked on the explainability of deep learning models, applied to computer vision and bioinformatics.
Interests
- Self-supervised learning
- Video world models
- Explainability and reasoning
Education
PhD in Deep Learning, 2018 - 2023
KTH - Royal Institute of Technology
MSc in Machine Learning, 2016 - 2018
KTH - Royal Institute of Technology
BSc in Computer Engineering, 2013 - 2016
UNIBO - University of Bologna
News
- Dec 2025 I’m joining Mistral in Paris as an AI Scientist 🇫🇷
- Nov 2025 Recognized as a top reviewer at NeurIPS 2025, thanks!
- Oct 2025 Guest lecture on DINOv3 to deep learning students at Chalmers University, Gothenburg, Sweden 🇸🇪
- Sept 2025 Honored to be invited to present DINOv3 at the Foundation Model workshop at the Geneva Institute of Theoretical Science 🇨🇭
- Sept 2025 Presenting my postdoc contributions (CAPI, dino.txt, DINO-world and DINOv3) to students of the WASP AI program at the University of Linköping, Sweden 🇸🇪
- Aug 2025 Announcing DINOv3, a major release that raises the bar of self-supervised vision foundation models 🦖🦖🦖
- Jul 2025 Publishing DINO-world, a latent video world model trained in the feature space of DINOv2. Also presented at ICML 2025 in Vancouver 🍁
- May 2025 Our work on scalable pre-training of dense image representations, CAPI, is accepted at TMLR. Congrats to Timothée for leading this work!
- Feb 2025 Our “DINO.txt” paper on aligning vision and language representations is accepted at CVPR 2025.
- Sept 2023 I’m moving to Paris to join the DINO team at Meta AI as a postdoctoral researcher 🦖
- Jun 2023 PhD Graduation at KTH, Stockholm 🎓🎉
- Feb 2023
Our work on semantic image compression with quantized autoencoders is accepted at ICASSP 2023 (Meta AI internship).
