AI Data Annotator (Freelance)
Performed 2D/3D joint annotation for autonomous driving, surveillance, retail, and voice datasets, focusing on high-volume throughput and quality. Applied semantic segmentation, bounding box and polygon labeling, keypoints labeling, and point cloud annotation while maintaining stringent accuracy targets. Served as a quality inspector and improved team performance through consistency checking and rule standardization. • Labeled 800+ boxes/day with 98.7% accuracy (top 5%). • Annotated 106-point facial keypoints with a 1.2% rejection rate. • Supported a 3-person team by raising accuracy from 94% to 97.5% via QA. • Created rule sheets that reduced onboarding time from 3 days to 1 day.