weloglobal - Welo Global - Multimedia Generative AI Analyst
Requirements
• Experience in video annotation, multimedia annotation, content quality review, data labeling, computer vision labeling, Generative AI evaluation, or a closely related field. • Strong attention to detail and ability to identify subtle errors, mismatches, and inconsistencies between prompts and videos. • Solid understanding of US driving rules, road behavior, traffic scenarios, and roadway conventions. • Good reading comprehension and ability to accurately compare written prompts against visual outputs. • Strong written communication skills in English, with the ability to describe errors clearly and concisely. • Ability to follow detailed guidelines consistently and maintain high accuracy across repetitive tasks. • Critical thinking and sound judgment when reviewing ambiguous or complex scenarios. • Basic understanding of robotics behavior in real-world environments. • Comfortable working in structured annotation platforms or similar tools. • Ability to maintain focus and quality while reviewing multimedia content for extended periods. • ## Ways to Stand Out from the Crowd • Prior experience evaluating AI-generated video, synthetic media, autonomous driving data, robotics scenarios, or traffic-related visual content. • Familiarity with US traffic rules, road signage, lane behavior, vehicle interactions, pedestrian behavior, and common driving scenarios. • Experience with QA, audit, or second-pass review workflows, including calibration, sampling, defect tracking, or error taxonomy development. • Ability to identify recurring model failure patterns across tasks or batches.
Responsibilities
• Review text prompts and corresponding AI-generated video clips to identify mismatches, inconsistencies, and visual errors. • Tag specific parts of the prompt that do not match the video by identifying only the incorrect prompt text, rather than the full sentence unless needed. • Add visual or behavioral errors as annotation instances in annotation platform. • Write brief, clear descriptions for each error instance. • Avoid duplicate annotations by ensuring errors already captured as prompt mismatches are not added again as separate instances. • Evaluate video content for actions, context, motion, scene consistency, object behavior, and prompt alignment. • Apply project guidelines consistently across repetitive, detail-oriented tasks. • Use critical thinking and judgment to handle ambiguous scenarios and determine the most accurate ground truth. • Identify recurring or systematic errors across tasks and document examples for review. • Perform self-QA on completed work and correct errors before submission. • Participate in calibration sessions to align interpretation of guidelines and reduce annotator-to-annotator variance. • Incorporate feedback from quality reviews into subsequent work. • Support throughput and quality targets while maintaining accuracy at scale.
Apply in one click
Upload My Resume
Drop here or click to browse · Tap to choose · PDF, DOCX, DOC, RTF, TXT