Video Gen Research Scientist - Scalable ML Pipelines

CANTINA RESEARCH SINGAPORE PTE. LTD.; Singapore; On-site; SGD 150,000 - 200,000; Full time

Benefits offered by this job

Competitive salary and generous equity

Personal time off and paid holidays

Health insurance

Global travel insurance

Monthly stipend: $500 (~S$635)

Home office equipment

Job summary

Cantina Research Singapore PTE. LTD. is seeking a Research Scientist to drive foundational research on video generation models and own the full research cycle from ideation to post-training improvements, collaborating with data, infrastructure, and modeling teams.

You will build scalable data systems for video data, design distributed pipelines, containerized workflows with Kubernetes, and optimize cloud storage across AWS, GCS, and Azure to deliver high-quality, cost-efficient model training.

Qualifications

• Experience building or scaling large-scale data systems for ML workflows.
• Strong background in post-training methods for large diffusion/flow-based models.
• Hands-on experience with distributed data processing and orchestration tools.

Responsibilities

• Build and maintain scalable systems for ingesting, preprocessing, and delivering large-scale video data for model training
• Design and scale distributed data pipelines for preprocessing, dataset generation, and repeated dataset refreshes
• Own workflow orchestration, job scheduling, monitoring, and failure recovery for large-scale data processing jobs
• Implement and maintain containerized pipeline infrastructure using Kubernetes or equivalent orchestration systems
• Optimize cloud-based data storage and movement across providers (AWS, GCS, or Azure) for cost, throughput, and operational efficiency
• Define and implement best practices for dataset storage layout, versioning, caching, retention, and access patterns
• Build tooling to support deduplication workflows at scale, including near-dedup pipelines over large video corpora
• Research and develop distillation methods for large-scale diffusion and flow-based video generation models, including guidance distillation and adversarial distillation
• Develop reward models and preference-based fine-tuning pipelines that align video generation quality with human judgments
• Analyze the relationship between base model behavior and post-training outcomes, and inform pretraining decisions accordingly

Skills

Large-scale data systems

PySpark

Ray

Airflow

Docker

Kubernetes

Python

PyTorch

Research publications

RLHF

Tools

FFmpeg

PyAV

DALI

OpenCV

Job description

Cantina Research Singapore PTE. LTD. is seeking a Research Scientist to drive foundational research on video generation models and own the full research cycle from ideation to post-training improvements, collaborating with data, infrastructure, and modeling teams.

You will build scalable data systems for video data, design distributed pipelines, containerized workflows with Kubernetes, and optimize cloud storage across AWS, GCS, and Azure to deliver high-quality, cost-efficient model training.