Project Based Robotics Annotation Rater

Innodata Inc.; Philippines; On-site; PHP 223,000 - 312,000; Full time

Job summary

Innodata Inc. Philippines is seeking a meticulous rater to review household-task videos, identify valid actions, and write clear Natural Language Descriptions (NLDs) following project guidelines.

AHT: Up to 60 minutes per job, depending on complexity and video length. Workflow includes reviewing the full video, selecting actions to include or exclude, and precisely tagging objects, actions, and hand designations.

Qualifications

• Strong English grammar and writing skills for accurate NLDs.
• Broad English vocabulary and familiarity with American household culture and objects.
• Excellent visual observation, including good color recognition and ability to notice subtle movements and object differences.
• High critical-thinking skills to understand nuances, edge cases, exceptions, and changing guidelines.
• Strong attention to detail for accurate segmentation, timestamps, and object/action identification.
• Strong reading comprehension for detailed and evolving guidelines.
• Learning agility and adaptability to quickly apply new instructions and feedback.
• Consistency and quality focus when handling repetitive and detailed tasks.

Responsibilities

• Review the full video and identify actions to include, exclude, group, or separate.
• Identify the correct object, action, hand designation, and destination when applicable.
• Write concise and grammatically correct Natural Language Descriptions (NLDs).
• Review work before submission and apply QPE feedback and guideline updates to succeeding tasks.

Skills

English grammar

Vocabulary

Visual observation

Critical thinking

Attention to detail

Reading comprehension

Learning agility

Consistency

Job description

Raters review household-task videos, identify valid actions, mark accurate start and end timestamps, identify accurate hand designation, and write clear Natural Language Descriptions (NLDs) based on project guidelines.

AHT: Up to 60 minutes per job, depending on complexity and video length.

Workflow

• Review the full video.
• Identify actions to include, exclude, group, or separate.
• Identify the correct object, action, hand designation, and destination when applicable.
• Write concise and grammatically correct NLDs.
• Review work before submission.
• Apply QPE feedback and guideline updates to succeeding tasks.

Required Skills

• Strong English grammar and writing skills for accurate NLDs.
• Broad English vocabulary and familiarity with American household culture and objects.
• Excellent visual observation, including good color recognition and ability to notice subtle movements and object differences.
• High critical-thinking skills to understand nuances, edge cases, exceptions, and changing guidelines.
• Strong attention to detail for accurate segmentation, timestamps, and object/action identification.
• Strong reading comprehension for detailed and evolving guidelines.
• Learning agility and adaptability to quickly apply new instructions and feedback.
• Consistency and quality focus when handling repetitive and detailed tasks.

Preferred Background

Experience in data annotation, AI training, video/image evaluation, QA, transcription, captioning, or similar guideline-based work is an advantage