World Model Research Intern (6-month Internship)

Tencent; Singapore; On-site; SGD 20,000 - 36,000; Full time

Job summary

Tencent in Singapore is seeking a World Model Research (Intern) to work on cutting-edge algorithm research and development for Game World Models, focusing on video generation and interaction.

You will assist in building automated data pipelines, develop efficient training frameworks for diffusion models, and collaborate with senior researchers to explore digital content generation and interactive agents in virtual worlds.

Qualifications

• Enrolled as a PhD or Masters student in Computer Vision, Computer Graphics, Machine Learning, or a related field, familiarity with generative models.
• Strong background in deep learning with practical experience in image or video data processing. Capable of independently handling complex data cleaning tasks and constructing robust data pipelines.
• Strong interest in applying generative AI to interactive entertainment and virtual world building.
• Good communication and collaboration skills.

Responsibilities

• Build automated game data collection pipeline to capture precise internal states and construct a high-quality training dataset.
• Help build highly efficient training frameworks for diffusion models, optimize training workflows, and take responsibility for managing, monitoring, and fine-tuning model training tasks.
• Deeply participate in technical project discussions, proactively propose innovative ideas or improvement plans, and actively drive the practical application and integration of research findings into game video generation and virtual world construction scenarios.
• Collaborate closely with senior researchers to explore cutting-edge directions in the fields of digital content generation and interactive agents.

Skills

Data processing

Deep learning

Communication

Collaboration

Education

PhD or Master\u2019s student in Computer Vision/Computer Graphics/ML

Tools

Generative models

Model training frameworks

Job description

Tencent is an Internet-based platform company founded in Shenzhen, China, in 1998. We use technology to enrich the lives of Internet users and assist the digital upgrade of enterprises. Our mission is "Value for Users, Tech for Good". We embrace a culture of teamwork & creativity and are driven by our values - Integrity, Proactivity, Collaboration and Creativity.

We are rapidly expanding our international operations and are looking for top talent to propel us forward. Combining the results-oriented nature of a start-up with the resources of a profitable and leading Internet company, Tencent offers a unique opportunity for aspiring individuals to thrive.

Responsibilities

Focus on cutting-edge algorithm research and development for Game World Models, exploring key issues such as video generation and video interaction. As a World Model Research (Intern), you will participate in one or more of the following tasks:

• Data Processing: Assist in building an automated game data collection pipeline to capture precise internal states and construct a high-quality training dataset.
• Model Training: Help build highly efficient training frameworks for diffusion models, optimize training workflows, and take responsibility for managing, monitoring, and fine-tuning model training tasks.
• Technical Innovation: Deeply participate in technical project discussions, proactively propose innovative ideas or improvement plans, and actively drive the practical application and integration of research findings into game video generation and virtual world construction scenarios.
• Frontier Exploration: Collaborate closely with senior researchers to explore cutting-edge directions in the fields of digital content generation and interactive agents.

Minimum Qualifications

• Enrolled as a PhD or Master’s student in Computer Vision, Computer Graphics, Machine Learning, or a related field, familiarity with generative models.
• Strong background in deep learning with practical experience in image or video data processing. Capable of independently handling complex data cleaning tasks and constructing robust data pipelines.
• Strong interest in applying generative AI to interactive entertainment and virtual world building.
• Good communication and collaboration skills.

Preferred Qualifications

• Model Expertise: Familiarity with mainstream video generation models; hands-on experience in training or fine-tuning these models is highly preferred.
• 3D Expertise: Familiarity with 3D-related research; experience in 3D modeling or generation is preferred.
• Academic Record: Experience publishing relevant research in top-tier AI and computer vision conferences or journals (e.g., CVPR, ICCV, ECCV, ICLR, ICML, NeurIPS) is a strong plus.