people

Currently a postdoctoral researcher at IRIT, working on Large Language Models within a European project: Geo-R2LLM

I was a postdoctoral researcher at LIPN (Laboratoire d’Informatique de Paris-Nord), where I work on machine learning for language and vision. My research interests sit at the intersection of computer vision, natural language processing, and interpretability / explainable AI (XAI), with a long-standing focus on human motion understanding.

I completed my PhD in Computer Science in 2024 at IMT Mines Ales (EuroMov Digital Health in Motion), advised by Andon Tchechmedjiev, Sylvie Ranwez, and Julien Lagarde. My thesis, “Attention for Human Motion Captioning: Towards an Interpretable Semantic Motion Segmentation and Analysis,” introduced attention-based architectures that generate language descriptions of human motion while remaining synchronized and interpretable — work that included winning the AffectMove challenge on protective-behaviour detection along the way. Parts of this research have since appeared at BMVC and AAAI, and in Neural Computing and Applications.

More recently, my work has extended into visual grounding for remote sensing imagery (MB-ORES) and probing large language models for geo-spatial knowledge, in collaboration with colleagues at LIPN and IRIT.

You can find my publications below, or browse them on Google Scholar and DBLP.