
Dvir Samuel
I hold a PhD from Bar-Ilan University, where I was supervised by Prof. Gal Chechik. My research focuses on long-tail and few-shot learning, as well as generative models, with a particular emphasis on diffusion- and flow-matching–based methods for image and video editing and personalization.
I began my academic journey in the “Mulla” program for gifted youth, which enables outstanding students to start their university studies during high school. Currently, I am a Senior Research Scientist at OriginAI, where I develop scalable and practical methods to advance Generative AI across images, videos, and 3D content.
Computer Vision
Long-tail Learning
Few-shot Learning
Selected & full publications