Hunyuan Multimodal Reinforcement Learning Research Intern

About the role#

This research internship focuses on advancing reinforcement learning (RL) algorithms for multimodal models. You will work within the Technology Engineering Group to develop strategies for image, video, and 3D generation, contributing to the development of next-generation RL paradigms that leverage environment feedback.

What you'll do#

• Conduct research on RL algorithms for multimodal models, including diffusion and autoregressive models.
• Design and develop RL infrastructure and reward modeling strategies to facilitate efficient large-scale training.
• Improve training stability and address challenges such as reward hacking.
• Explore new RL paradigms that learn directly from environment feedback.

What you'll need#

• Currently enrolled as a PhD student in Computer Science or a closely related field.
• Proven research capability with publications in top-tier conferences such as ICML, NeurIPS, ICLR, CVPR, ICCV, ECCV, or SIGGRAPH.
• Strong hands-on programming skills with experience in deep learning system implementation, model training, inference optimization, and distributed training.
• Prior experience with diffusion models, autoregressive models, or text-to-image/video generation is highly preferred.
• Participation in ACM or NOIP is considered a strong plus.

Location & details#

• Location: Palo Alto, California
• Modality: On-site
• Employment: Full-time internship
• Compensation: $80,168.40 to $124,800.00 per year
• Benefits: Eligible for paid sick leave, 13 paid holidays, and company-sponsored medical plan enrollment