Research Scientist, Video Gen & Scalable ML Systems

Cantina; Singapore; On-site; SGD 120,000 - 180,000; Full time

Benefits offered by this job

Salary + equity

Paid time off

Health insurance

Travel insurance

Monthly stipend

Home office equipment

Job summary

Cantina is expanding its Singapore team and seeks a Research Scientist to drive foundational research on video generation models. You will own data pipelines, distillation methods, and reward-based fine-tuning, collaborating with data, infra, and model teams to translate research into durable model improvements.

The role emphasizes scalable data systems, distributed processing, cloud storage optimization, and reproducible ML workflows across video, image, text, and audio modalities.

Qualifications

• Experience building or scaling large-scale data systems for ML workflows.
• Proficiency with distributed data processing frameworks (PySpark, Ray).
• Hands-on with containerization and orchestration (Docker, Kubernetes).
• Experience with cloud storage/compute (AWS, GCS, Azure).
• Familiarity with video/media tools (FFmpeg, OpenCV).
• Strong research background in post-training methods for diffusion/flow models.

Responsibilities

• Build and maintain scalable data systems for video model training.
• Design distributed data pipelines for preprocessing and dataset generation.
• Own workflow orchestration and job scheduling for large-scale data processing.
• Implement containerized pipeline infrastructure using Kubernetes or equivalent.
• Optimize cloud storage and data movement across providers for cost and performance.
• Develop distillation methods for large-scale video generation models.

Skills

Distributed data pipelines

PySpark / Ray

Docker / Kubernetes

Cloud storage (AWS/GCS/Azure)

Video processing (FFmpeg/OpenCV)

PyTorch / JAX

Research publications

Python programming

Tools

Airflow

Job description

Cantina is expanding its Singapore team and seeks a Research Scientist to drive foundational research on video generation models. You will own data pipelines, distillation methods, and reward-based fine-tuning, collaborating with data, infra, and model teams to translate research into durable model improvements.

The role emphasizes scalable data systems, distributed processing, cloud storage optimization, and reproducible ML workflows across video, image, text, and audio modalities.