Agentic Vision & Multimodal Reasoning Intern

Tencent International Service Pte. Ltd.; Singapore; On-site; SGD 150,000 - 210,000; Full time

Job summary

Tencent International Service Pte. Ltd. invites applications for a research-focused role in the Technology Engineering Group (TEG) in Singapore.

You will conduct R&D on Agentic Vision for Hunyuan multimodal foundation models, focusing on visual reasoning post-training and integrating perception, planning, tool execution, environment feedback, and self-correction. The role emphasizes advancing visual reasoning and spatial intelligence, developing multimodal post-training methods (SFT, PPO/GRPO,

Qualifications

• Master’s degree or higher in Computer Science, Artificial Intelligence, Computer Vision, Machine Learning, Mathematical Statistics, or related field.
• Strong foundations in deep learning and algorithm implementation, proficiency in PyTorch, and experience with distributed training.
• Experience with multimodal foundation models, visual reasoning, multimodal agents, or reinforcement learning for post-training.
• Familiarity with post-training methods (SFT, PPO/GRPO, OPD, RLVR) and reward modeling is a plus.

Responsibilities

• Conduct R&D on Agentic Vision for Hunyuan multimodal foundation models, focusing on visual reasoning post-training.
• Advance visual reasoning and spatial intelligence in complex scenarios, including 3D understanding and embodied planning.
• Develop multimodal post-training methods (SFT, PPO/GRPO, OPD) and verifiers/reward models.
• Build verifiable data synthesis pipelines using rendering engines and code sandboxes for data generation and testing.
• Create interactive sandboxes supporting code execution and visual tools to evaluate and improve models.

Skills

PyTorch

Distributed training

Multimodal reasoning

Deep learning

Research experience

Education

Master’s degree or higher in CS/AI/CV/ML/statistics

Tools

Megatron

OpenRLHF

RL frameworks

Job description

Tencent International Service Pte. Ltd. invites applications for a research-focused role in the Technology Engineering Group (TEG) in Singapore.

You will conduct R&D on Agentic Vision for Hunyuan multimodal foundation models, focusing on visual reasoning post-training and integrating perception, planning, tool execution, environment feedback, and self-correction. The role emphasizes advancing visual reasoning and spatial intelligence, developing multimodal post-training methods (SFT, PPO/GRPO,