Amélie Royer

Research Scientist · Kyutai Labs

I'm a Research Scientist at Kyutai Labs, a new AI lab committed to Open Science, based in Paris, France. My research interests are multimodal learning, computer vision, and neural network efficiency

Prior to this, I was a Deep Learning Research Engineer at Qualcomm AI Research (2021–2024), working on neural network efficiency via conditional compute and dynamic sparsity. I graduated with a PhD from IST Austria in 2020, working on Machine Learning and Computer Vision, and hold a Masters in CS from École normale supérieure de Rennes (2015).

📄

Selected Publications

Vision-Speech Models: Teaching Speech Models to Converse About Images

Amélie Royer*, Moritz Böhle*, Gabriel de Marmiesse, Laurent Mazaré, Neil Zeghidour, Alexandre Défossez, Patrix Pérez

Conference on Computer Vision and Pattern Recognition (CVPR) 2026

CASA: Cross-Attention over Self-Attention for Efficient Vision-Language Fusion

Moritz Böhle*, Amélie Royer*, Juliette Marrie*, Edouard Grave, Patrick Pérez

arXiv preprint 2025

MSViT: Dynamic Mixed-Scale Tokenization for Vision Transformers

Jakob Drachmann Havtorn*, Amélie Royer*, Tijmen Blankevoort, Babak Ehteshami Bejnordi

ICCV Workshop on New Ideas in Vision Transformers (NViT) 2023

Scalarization for Multi-Task and Multi-Domain Learning at Scale

Amélie Royer, Tijmen Blankevoort, Babak Ehteshami Bejnordi

Conference on Neural Information Processing Systems (NeurIPS) 2023