腾讯游戏-多模态大模型算法研究员-交互动画生成方向
Core
Researching multimodal algorithms for generating interactive 3D animations from text and video inputs, focusing on character-environment interactions and physical realism.
Role type
Senior IC multimodal AI researcher (interactive animation generation)
Builds
Real-time interactive 3D animations for game engines
Domain
Gaming + Computer Vision + Physics Simulation
Deliverable
production ML models
Required skills
Deep learning theory, multimodal fusion architectures (Single-stream/Double-stream), Transformer design, Python, PyTorch, 3D skeletal animation, human kinematics (IK/FK, SMPL), physics simulation (PhysX, Isaac Gym, MuJoCo), reinforcement learning for action control
Preferred skills
Publications in top conferences (CVPR, ICCV, ECCV, SIGGRAPH), open-source contributions, algorithmic competition wins
Technologies
PyTorch, PhysX, Isaac Gym, MuJoCo, SMPL
Responsibilities
Research multimodal animation generation algorithms involving text-to-motion and video-to-motion; Fuse geometric and contact perception with large model architectures; Research post-processing for physical simulation to ensure realism; Drive algorithm implementation in game engines to optimize inference efficiency for real-time generation
