Research Scientist, Video & Multimodal
$160k - $185k • Remote - United States
Posted 23d ago
Job Location
Remote - United States
Tech Stack
Remote Work Policy
On-site
Categories
AI Research Engineer
About the job
Innodata is seeking a Research Scientist to focus on the challenging area of video and multimodal AI models. This role involves partnering directly with customers and frontier labs to advance the science behind video understanding, video-language, and video-generation models. You will be instrumental in defining how data is designed, structured, and evaluated for these cutting-edge AI systems, ensuring that the data collected and the evaluations performed drive meaningful progress. Your work will directly influence the direction of research and development in this rapidly evolving field.
Responsibilities
- Translate video and multimodal model requirements into data specifications, including modalities, annotation schemas, sampling, and evaluation criteria.
- Develop evaluation methodologies for video understanding, focusing on temporal grounding accuracy, long-context reasoning, and cross-modal evaluation, determining when human judgment is necessary.
- Build evaluation methodologies for video generation, addressing fidelity, temporal coherence, and physical plausibility, especially where automatic metrics are insufficient.
- Determine how existing and incoming video data should be structured, enriched, and sampled to maximize model value.
- Conduct experiments to validate the impact of data decisions on model performance.
Requirements
- Experience in video understanding, video-language models, or video generation.
- Familiarity with multimodal AI models.
- Understanding of data specification, annotation schemas, and evaluation criteria.
- Experience in designing and executing experiments to validate data strategies.
- Ability to determine the effectiveness of model-based scoring versus human evaluation.