Research Scientist, Animation Direction, Level 4

Snap Inc is a technology company. We believe the camera presents the greatest opportunity to improve the way people live and communicate. Snap contributes to human progress by empowering people to express themselves, live in the moment, learn about the world, and have fun together. The Company’s three core products are Snapchat
, a visual messaging app that enhances your relationships with friends, family, and the world;
Lens Studio
, an augmented reality platform that powers AR across Snapchat and other services; and its AR glasses,
Spectacles
.

Snap Inc is a technology company. We believe the camera presents the greatest opportunity to improve the way people live and communicate. Snap contributes to human progress by empowering people to express themselves, live in the moment, learn about the world, and have fun together. The Company’s three core products are
Snapchat
, a visual messaging app that enhances your relationships with friends, family, and the world;
Lens Studio
, an augmented reality platform that powers AR across Snapchat and other services; and its AR glasses,
Spectacles
.

Snap Research serves as an innovation engine for the company. Our projects range from solutions to hard technical problems that significantly enhance Snap’s existing products, to riskier explorations that can lead to fundamental paradigm shifts in the way people communicate and express themselves. The team consists of scientists and engineers who experiment with and invent new technology that has a lasting impact on Snap’s products.

We are looking for a Research Scientist to join our Computational Imaging Research Team! The Computational Imaging team in NYC focuses on enhancing user experiences with our photo and short form video products, and enabling our creators to make Lenses more easily in Lens Studio. Specifically, our core focus in 2025 is on improving image and video quality (quality assessment, video super resolution, compression, image editing), and animations (animation generation, 3D avatar, video generation for animations).

Our research areas include computer vision, computational imaging/photography, 3d motion/video generation, efficient VLMs, and more.

What you'll do:

• Propose and develop cutting-edge multimodal techniques for 3D human animation and video generation — enabling expressive Bitmoji avatars and next-gen Snapchat video gen experiences

• Build models that map text, audio, video, and music to 3D animation and drive video generation downstream tasks, such as background video generation for animation rendering and audio-driven lip synced video generation

• Advance avatar technologies including animatable 3D Gaussian avatars, head/body reconstruction, and animation control

• Collaborate with product teams to deploy research at scale and publish at top-tier academic venues

Knowledge, Skills, & Abilities:

• Ability to define impactful research problems and deliver practical solutions, both in academic and product contexts

• Deep expertise in human motion modeling, facial/body animation, and generative modeling (e.g., text-to-motion, text-to-image, text-to-video, etc)

• Familiarity with video generation, avatar representations (e.g., mesh, NeRF, Gaussian), and rendering pipelines

• Strong prototyping and engineering skills (Python, PyTorch, C++), familiar with large scale distributed ML training on GCP/AWS clusters

• Proven ability to lead and mentor interns, PhD students, and junior researchers, as well as collaborate effectively with product teams

• Excellent communication and cross-functional collaboration skills

Minimum Qualifications:

• PhD in a technical field such as Computer Science/Electrical Engineering, or equivalent experience

• Hands-on experience with mainstream generative models (Diffusion, Transformers, GANs, VAEs) for animation or video

• Research or product experience in either of the following: 3D human motion generation from various signals such as text or audio; image/video generation; 3D avatar reconstruction and animation; or multimodal generation

Preferred Qualifications:

• Publications in top-tier venues as the main contributor (e.g., CVPR, SIGGRAPH, NeurIPS, ICCV, ECCV, ICLR); contributed to popular open-source…