Multimodal Reinforcement Learning Algorithm Researcher

About the role

Tencent's Technology Engineering Group (TEG) is hiring a Multimodal Reinforcement Learning Algorithm Researcher. TEG supports the company and its business groups on technology and operational platforms, runs R&D management and data centers, and operates the largest networking, device and data center network in Asia. The role focuses on research and development of reinforcement learning methods for multimodal models.

Key responsibilities

• Research reinforcement learning algorithms for multimodal models, including diffusion models for image and video generation, autoregressive models for multimodal understanding, and unified multimodal frameworks.
• Design and develop reinforcement learning training frameworks and reward modeling strategies to enable efficient large-scale training, improve stability and address issues such as reward hacking.
• Explore next-generation reinforcement learning paradigms that allow more direct and efficient learning from environmental feedback.

Required profile

• Bachelor's degree or above in Computer Science or a related field.
• Excellent research capability, with publications in top conferences such as ICML, NeurIPS, ICLR, CVPR, ICCV, ECCV and SIGGRAPH.
• Experience in diffusion models, autoregressive models, or text-to-image / text-to-video generation preferred.
• Participation in ACM/NOIP (Informatics Olympiad) preferred.

Required skills

• Strong engineering and programming skills.
• Deep learning system implementation, model training and inference optimization.
• CPU/GPU acceleration and distributed training and inference.