Research Scientist

• Drive foundational research on video generation models and post-training research
• 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 AWS, GCS, and Azure for cost, throughput, and operational efficiency
• Define best practices for dataset storage layout, versioning, caching, retention, and access patterns
• Build tooling for large-scale deduplication workflows, including near-deduplication pipelines over video corpora
• Research and develop distillation methods for diffusion and flow-based video generation models
• Develop reward models and preference-based fine-tuning pipelines aligned with human judgments on aesthetics, motion quality, and prompt adherence
• Analyze relationships between base model behavior and post-training outcomes
• Collaborate with data, infrastructure, adjacent modeling, and foundation model teams to translate research into durable model improvements and inform pretraining decisions

Explore Research Scientist at Cantina. Review requirements and application details.