Video Annotator - Bangladesh

We are looking for detail-oriented Video Annotators to support a Vision-Language-Action (VLA) dataset project. In this role, you will watch short video clips of physical tasks/activities and produce event-level annotations by segmenting each video into meaningful steps and writing clear, precise natural-language instructions describing the action taking place in each segment.

• Watch assigned videos in full before beginning annotation to understand the overall task and context.
• Segment each video into discrete, logical steps/events based on visible action boundaries.
• Write clear, concise, and grammatically correct step-by-step descriptions (instructions) for each segment, accurately reflecting what is happening on screen.
• Ensure annotations are action-oriented, unambiguous, and consistent in tense, tone, and phrasing across the dataset.
• Assign accurate start and end timestamps for each segment.
• Follow annotation guidelines, taxonomies, and style guides provided by the project team.
• Flag videos that are unclear, corrupted, mislabeled, or otherwise unsuitable for annotation.
• Participate in calibration sessions and incorporate reviewer/QA feedback to improve annotation quality and consistency.
• Meet daily/weekly productivity and quality targets without compromising accuracy.
• Maintain confidentiality of all project data and materials.