Video Annotator (Remote)

About the role

We are seeking detail‑oriented Video Annotators to support a Vision‑Language‑Action (VLA) dataset project. You will watch short video clips of physical tasks, segment them into meaningful steps, and write clear, concise instructions for each segment. The position is fully remote and can be engaged on a full‑time or contract basis.

Key responsibilities

• Watch each assigned video in full to understand the overall task before annotating.
• Segment videos into discrete, logical steps based on visible action boundaries.
• Write grammatically correct, action‑oriented step‑by‑step descriptions for each segment.
• Assign accurate start and end timestamps for every segment.
• Follow provided annotation guidelines, taxonomies, and style guides.
• Flag videos that are unclear, corrupted, or mislabeled.
• Participate in calibration sessions and incorporate reviewer feedback.
• Meet daily and weekly productivity and quality targets while maintaining data confidentiality.

Required profile

• English proficiency at C1 level (CEFR) with strong grammar and vocabulary.
• Excellent attention to detail and ability to notice subtle visual changes.
• Strong written communication skills for concise, precise descriptions.
• Ability to work independently, follow detailed guidelines, and maintain consistency.
• Reliable internet connection and a computer with a modern browser.

Required skills

• Experience with video annotation platforms such as CVAT.
• Familiarity with Label Studio.
• Knowledge of Scale or similar annotation tools.