AI Video Content Creator / RLHF Data Labeling & Evaluation
As an AI video content creator, I routinely performed comparative evaluations and RLHF-style judgments on outputs from multiple AI video generation models. My daily responsibilities included selecting superior AI-generated video outputs, identifying content quality, and detecting hallucinations, all to maximize user engagement and conversion. Over time, I developed systematic expertise in evaluating, rating, and red teaming AI outputs for quality and preference annotation. • Conducted hands-on testing and rapid cross-comparison of Sora, Kling, Runway, Pika, Jiemian, and Hailuo AI video models. • Performed A/B testing and RLHF-style preference annotation by selecting the best or most accurate model outputs. • Identified and documented edge cases, hallucinations, and model weaknesses for ongoing tool and workflow improvement. • Judged content engagement and market readiness, validating decisions through real-world performance metrics (e.g., 1M+ views and 1,000 sales).