David Pujol Perich

I am a PhD student at the Universitat de Barcelona, supervised by Sergio Escalera and Albert Clapés. My research lies at the intersection of Computer Vision and Natural Language Processing, with a focus on making video understanding models more robust and efficient, with publications at top-tier conferences like CVPR, ICCV, ICLR, WACV, etc.

During my Ph.D. I have investigated the domain gap between descriptive captions and realistic user search queries (CVPR 2026), developed methods for parameter-efficient fine-tuning (ICCV 2025), addressed visual domain shifts (WACV 2025). Moreover, I recently completed a research stay at the University of Bristol with Prof. Dima Damen and Prof. Michael Wray.

Previously, I earned my Master’s from Universitat Politècnica de Catalunya (UPC) and EPFL, where I worked on efficient attention at LIONS with Prof. Volkan Cevher. I also have a background in Graph Neural Networks resulting from working for 2 years at the Barcelona Neural Networking Center, where I engineered the IGNNITION framework (IEEE Network ‘21).

I am an invited lecturer for the Master of Computer Vision at UB and have received awards from SIGCOMM and the National Excellence Scholarship.