Video Sensitive Content Evaluation - VSCE
I managed the scope of Video Sensitive Content Evaluation (VSCE) projects for e2f Inc, auditing high-volume multimodal datasets to train AI safety models. Specific tasks included multi-labeling thousands of videos and titles across 15+ sensitive categories—such as sexual intent, physical violence, and gore—while performing independent source identification. Leveraging a professional background in Art Direction, I applied context-aware judgment to differentiate between cinematic expression and policy violations. Quality was maintained through strict adherence to zero-AI assistance protocols, rigorous timestamp-based viewing cycles (first, middle, and last 30 seconds), and consistent alignment with evolving safety taxonomies to ensure high-fidelity metadata accuracy.