At Creative Vision, we are pushing the boundaries of generative modeling across images, videos, 3D, and 4D. Our research is centered on a core challenge: enabling real-time, highly personalized interactive image and video models. We aim to unlock true generative rendering, allowing anyone in the world to instantaneously interact with generated objects and environments. Achieving this vision demands significant advancements in quality and efficiency. We train large foundational models to push the limits of quality and control. However, true interactivity requires immediate feedback, making efficiency paramount. This drives our heavy investment in optimizing streaming video models to run seamlessly, both on-device and on the backend. Over the years, our research has contributed to a number of experiences that millions of Snapchatters use every day.
Over the last 6 years our team has published 100+ of top conference papers. We organized highly attended tutorials based on our work: CVPR'21 Tutorial on Creativity, ECCV'22 Tutorial on Video Synthesis, CVPR’23 Tutorial on Efficient Neural Networks, CVPR’24 Tutorial on 3D/4D Generation and Modeling with Generative Priors. We have a robust internship and collaboration program, which in the past 6 years consisted of 80+ interns and collaborations with 20+ academic labs.
We have contributed to new experiences used by millions of Snapchatters daily: video generation, personalization, and 4D generation—all delivered with high efficiency.
In 2026 we’ll build even more ambitious projects, if our vision strikes a chord with you join us in 2026 as an intern. You’ll be working towards a publication in a top conference, participating in the larger effort of the team, training large foundational models that others are not even thinking about. We welcome both research and engineering interns.
Please email the team directly (preffered) or submit your application.
