The business problem involves advancing the development of large-scale video world models to maintain competitiveness and leadership in the field.
Requirements
- Solid foundation in deep learning algorithms and proven experience in large model R&D.
- Experience in Diffusion Models and Autoregressive Models, publications in top-tier conferences, or practical experience in text-to-image/text-to-video generation are preferred.
- Familiarity with implementation details of deep learning networks and operators, model tuning for training/inference, CPU/GPU acceleration, and distributed training/inference optimization.
- Participation in ACM/NOIP competitions is highly desirable.
Responsibilities
- Engage in the research and development of large-scale video world models, including the design and construction of training datasets, foundational model algorithm design, optimization related to pre-training, SFT, and RL, model capability evaluation, and exploration of downstream application scenarios.
- Analyze R&D challenges scientifically, identify performance bottlenecks, and develop solutions based on first principles to accelerate the development and iteration of world models.
- Explore diverse paradigms for world model implementation, research next-generation model architectures, and push the boundaries of world model capabilities.
Other
- Bachelor’s degree or higher (preferred) in Computer Science, Artificial Intelligence, Mathematics, or a related field.
- Strong learning, communication, and teamwork skills, coupled with a keen sense of curiosity.
