Applied Research Scientist, AI Research

Role overview Develop multimodal models that help an AI editing agent understand the visual and audio content of a media project, including the creative choices that shape an edit. The research focuses on practical capabilities for editing workflows, with work intended to move from experiments into production. You’ll help define how to represent and assess qualities such as whether a cut works or pacing feels right, where established evaluation approaches may not yet exist.

Responsibilities - Build vision-language systems that reason about visual and audio content in editing projects. - Design benchmarks and evaluations for editorial quality, balancing output quality with cost and latency. - Create the datasets needed for model development, including synthetic examples when real data is scarce. - Train specialized models from scratch or adapt existing foundation models to meet product needs. - Move research prototypes into production in collaboration with agent and engineering teams. - Help set research direction for multimodal understanding in creative editing workflows.

Requirements - The available source does not provide specific qualifications or experience requirements for this position.

Benefits and work setup - Based in San Francisco or remote within the United States.