About the role#
The Seed Multimodal Interaction and World Model team develops models with human-level interaction capabilities and multimodal understanding. This internship offers the chance to contribute to research and product development. You will work alongside industry experts on emerging technologies.
What you'll need#
- Currently pursuing a PhD in Computer Science, Computer Engineering, Software Development, or a related technical field.
- Research background in reinforcement learning, multimodal learning, video understanding, or vision-language modeling.
- Publications in venues like CVPR, ECCV, ICCV, NeurIPS, ICLR, or ICML.
- Ability to obtain and maintain work authorization in the United States.
- Experience with reinforcement learning in interactive environments is preferred.
- Familiarity with video generation or diffusion-based generative models is a plus.
- Experience with large-scale model training, such as distributed training, curriculum learning, or memory-augmented transformers, is preferred.
- Strong programming skills and experience building training or evaluation pipelines for machine learning models.
What you'll do#
- Design and implement reinforcement learning training systems for large-scale multimodal foundation models.
- Build unified modeling frameworks that integrate video, audio, and language, focusing on visual latent reasoning.
- Explore reinforcement learning approaches to bridge understanding and generation for multimodal visual reasoning.
- Work with researchers to evaluate models on world modeling, reasoning, and instruction-conditioned generation tasks.
Location & details#
- Location: San Jose, California
- Modality: On-site
- Type: Full-time internship
- Compensation: $85 per hour
- Sponsorship: Not provided


