游戏AI-多模态世界模型算法专家-可控视频生成方向
Core
Research and develop world model algorithms for game scenarios, focusing on controllable video generation, prediction, and editing.
Role type
Senior IC machine-learning engineer (video generation & world models)
Builds
Interactive, editable, physically and spatiotemporally consistent world models for games
Domain
Gaming + Generative AI
Deliverable
production ML models
Required skills
PyTorch, Diffusion Models, GANs, VAEs, Flow-based Models, multi-modal large models (CLIP, LLaVA, Qwen), large-scale GPU cluster training, fine-tuning, post-training
Preferred skills
First-author publications in SIGGRAPH, CVPR, ICCV, NeurIPS
Technologies
PyTorch, CLIP, LLaVA, Qwen
Responsibilities
Explore frontier technologies in multi-modal understanding, generative AI, and reinforcement learning; Develop and deploy core world model algorithms for video generation, prediction, and editing; Build next-generation video generation technologies for game worlds with interactivity and physical consistency
