Research Intern — Generative Video AI (Onsite Singapore)

Speedrun Talent Network; Singapore; On-site; SGD 17,000 - 23,000; Full time

Benefits offered by this job

Visa and travel support for eligible国际

Housing support in Singapore

Research stipend

Job summary

Cantina Labs is seeking exceptional research interns to join its Singapore lab for a three-month onsite internship beginning October 2026. You will work on next-generation video models under a senior mentor, with ownership over a well-defined research or engineering problem.

Projects span post-training efficiency, reward modeling, multimodal data systems, and scalable training infrastructure. You will present findings, contribute to publications, and receive a competitive stipend with visa

Qualifications

• Pursuing a PhD or final-year Master’s in CS/ML/CV or related field.
• Hands-on experience with diffusion and flow-based video generation models.
• Experience with post-training techniques such as distillation and reinforcement learning.
• Ability to formulate hypotheses, design controlled experiments, and analyze results.
• Proficient in Python; experience with PyTorch and/or JAX is a plus.

Responsibilities

• Research and develop distillation methods for large-scale diffusion and flow-based video generation models.
• Explore techniques to reduce inference cost while maintaining or improving generation quality.
• Develop reward models and preference-based optimization to improve aesthetics and timing.
• Study how base-model behavior affects post-training outcomes and inform model development.
• Design rigorous evaluations and conduct large-scale experiments on generative video models.
• Contribute to evaluation tooling, model integrations, or product-adjacent projects.

Skills

Python

Diffusion models

Flow-based models

Reinforcement learning

Preference optimization

research experience

PyTorch

JAX

Education

PhD candidate or final-year Master’s in CS/ML/CV

Tools

PyTorch

JAX

Job description

Cantina Labs is seeking exceptional research interns to join its Singapore lab for a three-month onsite internship beginning October 2026. You will work on next-generation video models under a senior mentor, with ownership over a well-defined research or engineering problem.

Projects span post-training efficiency, reward modeling, multimodal data systems, and scalable training infrastructure. You will present findings, contribute to publications, and receive a competitive stipend with visa