VLA Video Annotator: Segmentation & Captioning Expert

CNTXT AI; Philippines; Remote; PHP 240,000 - 360,000; Full time

Job summary

CNTXT AI is seeking detail-oriented VLA Annotators in the Philippines to support high-quality training data for Vision-Language-Action and robotics models. You will review videos, segment actions with precise timing, and craft clear descriptions of on-screen activities.

You will validate AI-generated captions and ensure consistency and accuracy across annotations. Ideal candidates have strong attention to detail, excellent written English, and the ability to follow complex guidelines.

Qualifications

• Strong attention to detail and ability to follow detailed written guidelines.
• Excellent written English with clear, concise action descriptions.
• High observational accuracy to distinguish small differences in actions and timing.
• Experience reviewing and editing video annotations or captions preferred.
• Comfort with repetitive tasks while maintaining accuracy and consistency.

Responsibilities

• Review videos and understand the task or activity being performed.
• Divide videos into action-based segments with accurate start/end times.
• Ensure all meaningful actions are captured without gaps or overlaps.
• Review and edit AI-generated captions for accuracy and clarity.
• Describe actions using simple, precise language based on visible evidence.
• Highlight relevant hands, objects, tools, and locations when needed.
• Maintain consistent terminology throughout each annotation.
• Identify invalid or unusual portions of videos per guidelines.
• Review completed work for segmentation, caption accuracy, grammar, and completeness.

Skills

Attention to detail

Guidelines comprehension

Strong written English

Observational skills

Video review

Repetitive annotation

Computer proficiency

Independence

Reliability

Feedback openness

Tools

Annotation platforms

AI-assisted labeling tools

Job description

CNTXT AI is seeking detail-oriented VLA Annotators in the Philippines to support high-quality training data for Vision-Language-Action and robotics models. You will review videos, segment actions with precise timing, and craft clear descriptions of on-screen activities.

You will validate AI-generated captions and ensure consistency and accuracy across annotations. Ideal candidates have strong attention to detail, excellent written English, and the ability to follow complex guidelines.