Nedko Savov

INSAIT, PhD Student

Nedko Savov is a PhD student at INSAIT, Sofia University, supervised by Prof. Luc Van Gool. His research focuses on generative video and interactive world models, with interests in long-context processing, memory, controllability, and generalization across environments.

His work explores model architectures, world-state representations, and memory mechanisms that can support persistent, large-scale interactive generative worlds, as well as methods for evaluating their capabilities. More broadly, he is interested in controllable visual generation, learned action representations, generating long videos efficiently, and building world models that generalize across many environment dynamics and actions.

His work has appeared at NeurIPS, CVPR, ICLR, and ECCV. He has also co-organized the NeurIPS 2025 Workshop on Embodied World Models for Decision Making.

Previously, he worked as a scientific programmer, with a background in face analysis and synthesis, 3D vision, and computational photography. He holds an M.Sc. in Artificial Intelligence from the University of Amsterdam and a B.Sc. in Computer Science from Sofia University.