Research Intern: Efficient Video Diffusion & Proxies

XG TECH PTE. LTD.; Singapore; On-site; SGD 20,000 - 31,000; Part time

Job summary

XG Tech PTE. LTD. in Singapore is seeking a Research Intern to explore efficient and controllable video generation for interactive applications. You will work with mentors to develop models, run experiments, and contribute to technical reports and publications.

Ideal candidates are pursuing a PhD in CS/AI/CV or a strong Master’s student with relevant research experience. Proficiency in Python and PyTorch, plus a background in diffusion models or generative video approaches, is required.

Qualifications

• Pursuing PhD or exceptional Master’s in CS/AI/CV.
• Strong foundations in deep learning; familiarity with diffusion models.
• Experience in generative modelling, video generation, or model efficiency.
• Ability to read research papers, reproduce methods, design experiments, and analyze results.

Responsibilities

• Investigate video generation methods conditioned on structured visual proxies, with control over scene layout, object motion, and camera movement.
• Explore diffusion acceleration techniques such as few-step distillation, adaptive sampling, feature caching, and efficient attention.
• Research ways to use scene and motion information to reduce redundant computation while preserving visual quality and temporal consistency.
• Implement and evaluate models, reproduce relevant research, and conduct controlled experiments.
• Support the development of paired proxy–video datasets and training pipelines.
• Evaluate generation quality, control accuracy, inference latency, and GPU memory usage.
• Document findings and contribute to research discussions, technical reports, and potential publications.

Skills

Python

Deep learning

Diffusion models

Research experience

Experiment design

GPU performance optimization

Reading papers

Education

PhD or exceptional Master’s in CS/AI/CV

Tools

PyTorch

Job description

XG Tech PTE. LTD. in Singapore is seeking a Research Intern to explore efficient and controllable video generation for interactive applications. You will work with mentors to develop models, run experiments, and contribute to technical reports and publications.

Ideal candidates are pursuing a PhD in CS/AI/CV or a strong Master’s student with relevant research experience. Proficiency in Python and PyTorch, plus a background in diffusion models or generative video approaches, is required.