Robotics Video Annotation Expert

The work

You will review short videos of robots or people performing actions. You will decide whether each action is accurate, complete, and consistent with written instructions, defined end states, and project guidelines.

You will also compare video samples, improve AI-generated annotations, and record clear reasons for your decisions. The work requires careful attention to movement, object interaction, action boundaries, and the order of actions.

• Review and analyze short videos involving robots or human activity.
• Check whether actions follow written directives and reach the required end state.
• Compare samples and identify which one better fulfills the instruction.
• Correct or add action boundaries, descriptions, and feedback to AI-generated annotations.
• Document observations and decisions in clear written language.
• Apply detailed rubrics consistently and maintain annotation quality.

What it pays and takes

This is an entry-level contractor project. Prior AI experience is not required, but you need to make careful visual judgments and apply detailed instructions consistently.

• Pay: $15-$80 USD per hour.
• Project length: approximately 8 weeks.
• Workload: part time, with an estimated 20 hours per week and a listing requirement of 20+ hours per week.
• Location: remote, with eligibility limited to the countries listed for this project.
• Language: English.
• Useful background: robotics, human activity, video annotation, or quality assurance.
• You should have strong attention to detail, sound judgment when comparing samples, precise editing skills, clear written communication, and consistent quality-assurance habits.

How it works

Apply on OpenTrain with your resume and then complete the application on the hiring site.

About AI training work

AI training is the human work behind systems that learn from examples. People review videos, text, audio, and other data so AI can recognize actions and produce more useful results; experience in a subject area can help workers make more accurate judgments.