Vicky Kalogeiton is a Professor in AI at the Computer Science Laboratory (LIX) of École polytechnique, where she leads the VISTA team. She is an ELLIS member attached to the Paris Unit, a contributing researcher at Archimedes, and a Hi! PARIS Fellow and a member of its scientific committee.

Her research focuses on multimodal generative AI, with a particular interest in efficiency, structured and multiple outputs, and medical applications. She develops generalizable methods that can be applied across domains and publishes in leading computer vision conferences and journals, including CVPR, ICCV, ECCV, T-PAMI, and IJCV. Before joining École polytechnique, she was a research fellow at the Visual Geometry Group at the University of Oxford. She completed her PhD at the University of Edinburgh and INRIA Grenoble, under the supervision of Vittorio Ferrari and Cordelia Schmid.

Her Hi! PARIS fellowship project focuses on developing more flexible and robust AI models for Earth Observation data.

It aims to build a multimodal geospatial foundation model capable of working with diverse data sources, including satellite imagery, radar time series, LiDAR point clouds, hyperspectral images, very high-resolution photographs, and historical maps. The goal is to better address specialized environmental and geospatial challenges that current models often fail to capture, such as biomass estimation, land cover mapping, crop analysis, archaeological site detection, and historical image interpretation.

By combining self-supervised learning, multimodal representation learning, and model distillation, the project seeks to make geospatial AI more adaptable, efficient, and accessible for real-world applications.