Research Scientist - 3D Human Pose and Shape Estimation

Meta Reality Labs Research (RL-R) is a world leader in the design of virtual and augmented reality systems. We are currently seeking innovative researchers with a passion for technology to develop the next generation of digital humans and avatars. The focus of this role is on human motion and appearance reconstruction. The position requires a combination of education and experience in computer vision, computer graphics, physics simulation and deep learning.

Research Scientist - 3D Human Pose and Shape Estimation Responsibilities:

• Conceive, plan, and execute cutting-edge research programs to advance the state-of-the-art in digital humans and avatars.
• Formulate and develop new advanced technologies related to the generation of convincing digital humans, in areas such as dense dynamic 3D surface reconstruction, statistical deformation models, human biomechanics, motion synthesis and retargeting, markerless motion capture or physics-based simulation.
• Collaborate with other researchers, engineers, and 3D artists to prototype hardware and software systems and solutions, experiments and concepts that advance the entire project.
• Communicate complex research topics to an audience of diverse backgrounds, and be able to explain the importance and relevance of long term research agenda.

Minimum Qualifications:

• PhD in related fields such as Computer Graphics/Vision, Machine Learning, Robotics, Computer Science, Applied Math, or Physics with a proven academic track-record.
• First-authored publications at leading conferences or journals such as SIGGRAPH (North America or Asia)/TOG, CVPR, ECCV/ICCV, NeurIPS, RSS.
• Proficient in C++ or Python programming languages.
• Strong understanding in one or more of the following topics: multivariate calculus, linear algebra, kinematics of articulated structures, 3D data representation and spatial acceleration structures, camera calibration, SLAM and fusion, numerical optimization, optimal control, Newtonian mechanics, and numerical methods for solving differential equations.

Preferred Qualifications:

• 3+ years’ experience in tracking and reconstruction of realistic digital humans, including deformable geometry tracking and reconstruction, anatomical motion modeling, and physically-based simulation and animation.
• Proven track record of achieving significant results as demonstrated by grants, fellowships, or patents.
• Proven track record of leading research projects involving 5+ people to completion.
• Experience working with modern deep learning frameworks and neural network architectures.
• Experience with cuda kernel development or neural network model optimization.
• Experience with differentiable physics simulation.
• Experience with physics-based rendering, real time rendering, differentiable rendering, or neural rendering.
• Experience contributing to open source projects or datasets.
• Experience working with high-end VFX content creation software and/or video game engines.