On 3–4 July 2025, the DVPS project kicked-off in Rome, bringing together 70 leading researchers from 20 institutions across 9 European countries. Over the next four years, DVPS will push the boundaries of artificial intelligence by developing a new generation of Multimodal Foundation Models (MMFM)—AI systems capable of understanding and integrating not just text, but also vision, sound, and real-world physical context.

The potential applications are vast: from language and communication, to healthcare (cardiology), to environmental and geo-intelligence. DVPS aims to make AI more adaptable, inclusive, and powerful—by teaching it to learn the way humans do: through interaction and context.

🌍 Kick-off in Rome
The project’s official start was marked by a two-day meeting filled with energy, collaboration, and a shared sense of purpose. Representing BCN-AIM, Karim Lekadir and Cristian Izquierdo joined project partners in laying the groundwork for this ambitious initiative. Our team will lead efforts on federated learning, enabling secure, privacy-respecting AI training across institutions.

We’re proud to be part of this groundbreaking consortium and excited to share future milestones as the project unfolds.

🔗 Learn more about DVPS: https://dvps.translated.com/