Research Scientist, Video Foundation Models

What you'll do

• Research and scale video and multimodal models through pre-training, continued training, and post-training.
• Develop architectures, objectives, conditioning methods, and pipelines for generation, multimodal interaction, and audio-video generation.
• Run experiments on model quality, controllability, consistency, robustness, and efficiency; collaborate on technical direction and research outputs.

What they ask for

• Strong research and engineering experience in generative modeling, such as diffusion models, flow matching, DiTs, video generation, or 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.
• Track record of impactful models or systems, shown through publications, open-source work, production impact, or other significant technical work.
• Ability to own ambiguous research problems and specialized depth in one or more foundation-model lifecycle areas.

Pay

$200,000-$320,000 USD annual base salary

Tools

diffusion models, flow matching, DiTs

AI focus

Research and develop native video and multimodal foundation models for generation and understanding, including reference- and memory-based generation, multimodal interaction, and joint audio-video generation.