Vision-Language-Action (VLA) Annotator - Kenya

Vision-Language-Action (VLA) Annotator

Location: Remote

Type: Full-time / Contract

About the role

We are looking for detail-oriented VLA Annotators to support the creation of high-quality training data for Vision-Language-Action and robotics models.

Annotators will review videos of people performing real-world activities, divide them into meaningful action segments, and create or validate clear descriptions of the actions taking place. The role also involves reviewing AI-generated annotations, correcting inaccuracies, and ensuring the final data meets project-specific quality standards.

Projects may include household, workplace, industrial, manipulation, navigation, and other real-world activities.

Responsibilities

• Review videos and understand the overall task or activity being performed.
• Divide videos into logical action-based segments with accurate start and end times.
• Ensure all meaningful actions are captured without unnecessary gaps or overlaps.
• Review AI/VLA-generated captions and select, edit, or replace them when necessary.
• Write clear and concise action descriptions based only on what is visibly happening in the video.
• Accurately describe relevant hands, objects, tools, movements, and locations when required.
• Maintain consistent terminology for objects and actions throughout each annotation.
• Identify invalid, idle, distracted, failed, recovery, or otherwise unusual portions of a video according to project guidelines.
• Review completed annotations for segmentation, caption accuracy, grammar, consistency, and completeness before submission.
• Follow project-specific annotation guidelines and adapt quickly when requirements change.
• Flag ambiguous examples, recurring model errors, edge cases, or unclear instructions to the Operations/Quality team.
• Use annotation shortcuts and AI-assisted tools efficiently while maintaining required quality and productivity targets.
• Participate in calibration, training, and feedback sessions as needed.

Requirements

• Strong attention to detail.
• Ability to understand and consistently apply detailed written guidelines.
• Good written English and the ability to describe actions clearly using simple, precise language.
• Strong observational skills and ability to distinguish small differences in actions, objects, and timing.
• Comfortable reviewing video content for extended periods.
• Ability to perform repetitive annotation work while maintaining accuracy and consistency.
• Basic computer proficiency and ability to learn new annotation platforms quickly.
• Ability to work independently and meet defined quality and productivity targets.
• Reliable computer and internet connection.
• Openness to feedback and ability to quickly incorporate guideline changes.

Nice to have

• Previous experience in data annotation, video annotation, quality assurance, robotics data, computer vision, or AI training data.
• Experience working with temporal segmentation or action recognition tasks.
• Familiarity with AI-assisted labeling tools or reviewing model-generated outputs.
• Experience with egocentric / first-person video data.
• Basic familiarity with robotics, manufacturing, household tasks, or other physical-world activities.

What success looks like

• Accurate segmentation of real-world actions.
• Clear, self-contained, and consistent captions.
• Minimal missed actions, unnecessary segments, or timing errors.
• Strong judgment when reviewing AI-generated annotations.
• Consistent quality across large volumes of video.
• Ability to identify recurring annotation or model issues rather than treating every task in isolation.