混元AIGC算法研究员（世界模型基模方向）（北京/上海）
Core
Research and development of video world foundation models, focusing on training data design, algorithm development, and optimization for pre-training/SFT/RL.
Role type
Senior IC research scientist (world models/foundation models)
Builds
Video world foundation models and downstream application scenarios
Domain
Artificial Intelligence / Generative AI / Video
Deliverable
production ML models
Required skills
Deep learning algorithms, Large model development, Diffusion Models, Autoregressive Models, Model architecture design, Training and inference optimization, Distributed training
Preferred skills
Top-tier paper publications, Text-to-image/video generation experience, ACM/NOIP competition experience, Low-level operator implementation
Technologies
Diffusion Models, Autoregressive Models, GPU/CPU acceleration
Responsibilities
Design and construct training data for video world models; Optimize pre-training, SFT, and RL processes; Explore next-generation model architectures and boundaries; Analyze performance bottlenecks and implement solutions from first principles.
