Honors
- 2026Gold Reviewer Award, ICML 2026
- 2025Spotlight paper, NeurIPS 2025 (Alligat0R)
- 2023PhD funded by the Bosch Research Foundation
About
I am a PhD candidate at IMAGINE (LIGM, École des Ponts, IP Paris), advised by Vincent Lepetit and Guillaume Bourmaud. My thesis, Automatic Semantic Scene Learning for Robust and Accurate Localization, is funded by the Bosch Research Foundation.
My research is about learning 3D-aware visual representations from image pairs and videos. I work on pre-training for 3D vision, from co-visibility supervision (Alligat0R) to self-supervised objectives (Gekko, Poincar3), and on understanding and benchmarking image matching (RUBIK, RoMa-Ω). I have also been a visiting researcher at the Visual Geometry Group, University of Oxford.
Before my PhD, I graduated from the MVA Master (Mathématiques, Vision, Apprentissage) at ENS Paris-Saclay, after working as an IT consultant at Deloitte. I also hold an engineering degree from CentraleSupélec and a master's degree in Nanoscience from Université Paris-Saclay.
News
-
Preprint of our new work Poincar3: Emergent Multi-View Geometry Through Self-Distillation is now available on arXiv, along with the project page and code and weights.
-
Our paper Gekko has been accepted at NeurIPS 2026!
-
I attended ECCV 2026 in Malmö, where I presented Gekko at the UniWorld workshop and RoMa-Ω at the SfM-ADL workshop.
-
Our paper RoMa-Ω: What Feed-Forward 3D Models Know About Image Matching was accepted at the SfM-ADL workshop at ECCV 2026. Code is available.
-
Preprint of our work Gekko is now available on arXiv, along with the project page, code and pre-trained models.
-
I received an ICML 2026 Gold Reviewer Award!
-
I presented my work at Bosch.
-
I started a visiting period at VGG, University of Oxford.
-
I attended NeurIPS 2025 in San Diego and visited the BAIR Lab at UC Berkeley.
-
I gave a talk and presented a poster at NeurIPS in Paris, a local alternative to the NeurIPS conference.
-
Our work Alligat0R has been accepted to NeurIPS 2025 as a Spotlight!
-
I attended ICVSS 2025 in Sicily!
-
Preprint of our work Alligat0R is now available on arXiv.
-
Our paper RUBIK has been accepted to CVPR 2025!
-
Our work on Reliability in Semantic Segmentation was published at ECCV 2024!
03
Publications
Experience
Research & work
-
2023 – now
PhD Candidate
IMAGINE, École des Ponts · Champs-sur-Marne, France
Pre-training and self-supervised learning for 3D vision, camera pose estimation and image matching.
-
2026
Visiting Researcher
Visual Geometry Group (VGG), University of Oxford · UK
-
2023
Research Intern
Valeo.ai · Paris, France
Generative synthetic data to assess the reliability of semantic segmentation models (ECCV 2024).
-
2021 – 2022
IT Consultant
Deloitte · Paris, France
-
2021
Research Intern
C2N, CNRS & Université Paris-Saclay · Palaiseau, France
Energy-efficient deep learning on spintronic (STT-MRAM) hardware.
Education
-
2023 – 2027
PhD in Computer Vision
École des Ponts (ENPC), Institut Polytechnique de Paris
Expected defense: early 2027. Advised by Vincent Lepetit and Guillaume Bourmaud.
-
2022 – 2023
MSc MVA: Mathematics, Vision, Learning
ENS Paris-Saclay
-
2020 – 2021
MSc in Nanoscience
Université Paris-Saclay
-
2017 – 2021
Engineering Degree (Diplôme d'Ingénieur)
CentraleSupélec
Major in Communicating Systems & Internet of Things.
Honors, talks & service
Talks & visits
- 2026Visiting researcher, VGG, University of Oxford
- 2026Research presentation, Bosch
- 2025Research visit, BAIR, UC Berkeley
- 2025Talk & poster, NeurIPS in Paris
- 2025ICVSS summer school, Sicily
Teaching & service
- 2025Introduction to C++ (lectures & labs), École des Ponts
Reviewer for
- CVPR
- ECCV
- ICLR
- ICML
- NeurIPS
- IJCV