Meta AR/VR Job | Research Scientist – Neural Rendering
Type（岗位类型）: 3D Software Engineering | Artificial Intelligence, Computer Vision, Games, Machine Learning, Research
Citys（岗位城市）: Sausalito, CA
For AR/VR communication to be a success, it has to continue the tradition of phone and video calling: you’re hearing and seeing what the other person sounds and looks like. Our goal is to create a truly photographic experience for communication. For this, we are exploring ways to render a digital representation of a user that is high fidelity and highly realistic. And to do this Facebook is exploring completely new ways of thinking about rendering.
In this position, we want to research and explore using machine learning neural rendering techniques to work in conjunction with the latest technology in rendering to produce a digital representation of a user that is as immediate and accurate as a video of a user. The ideal candidate will have a strong understanding of computer graphics and machine learning techniques with an emphasis on image-to-image translation.
5+ years of work experience working with game engines in the games or movie industry
Experience working with shading pipelines
Solid understanding of the rendering and shading stack of game engines
Solid understanding of multi-process and GPU architectures
A broad understanding of machine learning techniques
A broad understanding of challenges in rendering, in particular for AR/VR applications
Communication and organizational skills
Demonstrated experience working with minimal supervision to complete tasks
Exhibiting listening and communication skills, experience receiving direction and criticism
Demonstrated experience working collaboratively within a team environment
Experience picking up new techniques and experience learning new software quickly and easily
Experience working with and expanding the usability of new research software
Currently has, or is in the process of obtaining a Bachelor’s degree in Computer Science, Computer Engineering, relevant technical field, or equivalent practical experience. Degree must be completed prior to joining Meta.
Develop new ways of rendering the head, hair and body of a user in a highly photographic manner
Research techniques for combining image-to-image translation and deferred neural rendering with highly accurate traditional rendering techniques
Build prototypes and demos of real-time image-to-image enhanced digital humans
Research disentangling the components of an image: lighting, pose, expression and identity
Research techniques to quickly create 3D real-time representations of the user as they are in the moment
Experience with some high level shading languages (RSL, OSL, HLSL, AZSL), Python, CUDA C, C++, PyTorch