3D生成算法工程师（数字人动捕方向） - 抖音直播
Core
Develop algorithms for digital human motion capture and driving in live streaming scenarios, focusing on full-body pose estimation, hand tracking, and facial expression capture.
Role type
Senior IC 3D generation algorithm engineer (motion capture)
Builds
Real-time motion capture systems for virtual streamers
Domain
Live streaming, Computer Vision, 3D Graphics
Deliverable
production ML models
Required skills
Deep learning, Computer vision, PyTorch, Human pose estimation, Hand keypoint detection, Facial expression capture (Blendshape/4D Face), Audio2Face, Human motion generation, Motion capture data processing (BVH/FBX), Model compression optimization
Preferred skills
Mobile real-time inference, CVPR/ICCV/ECCV/SIGGRAPH publications, Virtual streamer/industry experience
Responsibilities
Research and develop algorithms for digital human motion capture and driving; Develop monocular video motion capture algorithms for high-precision, low-latency real-time tracking; Optimize inference speed and resource usage for mobile/PC real-time streaming; Collaborate with product and engine teams to deploy motion capture capabilities in virtual streamer scenarios.
