Research Scientist, Multimodal Video AI

Cantina; California (MO); On-site; USD 200,000 - 320,000; Full time

Benefits offered by this job

Competitive salary

Equity

Medical / Dental / Vision

Paid time off

Parental leave

401(k)

Lifestyle stipend

In-office meals

One Medical

Job summary

Cantina Labs is building a core team to advance native video and omni foundation models for multimodal generation and understanding. The role covers pre-training, continued training, and post-training for high-quality, controllable generation.

You will influence architecture, data, training, evaluation, and efficient inference, shaping technical direction early on. You will collaborate with researchers, engineers, and product teams, with opportunities to publish research and contribute to

Qualifications

• Strong research and engineering experience in generative modeling, including diffusion models, video generation, and multimodal models.
• Hands-on experience training and evaluating large-scale image, video, or unified multimodal models using modern deep learning frameworks and distributed training systems.
• A track record of developing impactful models or systems, demonstrated through publications, open-source contributions, or production impact.
• Ability to independently own ambiguous research problems and move from ideas to experiments in a highly collaborative environment.
• Specialized depth in foundation model lifecycle areas such as architecture, data curation, controllable generation, or post-training.

Responsibilities

• Research, develop, and scale native video and multimodal foundation models from prototypes to large-scale pre-training, continued training, and post-training.
• Explore new architectures, training objectives, and conditioning mechanisms for video generation and multimodal interaction.
• Build and improve data curation, distributed training, evaluation, and post-training pipelines for high-quality generation.
• Design systematic experiments to understand model scaling, generation quality, controllability, and inference efficiency.
• Collaborate with researchers, engineers, and product teams to shape roadmaps and translate advances into capabilities.
• Contribute to research publications and open-source releases when appropriate.

Job description

Cantina Labs is building a core team to advance native video and omni foundation models for multimodal generation and understanding. The role covers pre-training, continued training, and post-training for high-quality, controllable generation.

You will influence architecture, data, training, evaluation, and efficient inference, shaping technical direction early on. You will collaborate with researchers, engineers, and product teams, with opportunities to publish research and contribute to