Data Annotation Specialist / AI Trainer
Delivered high-quality computer vision training data through annotation, QA, and relabeling workflows for large-scale projects. • Produced and validated 240,000+ bounding boxes, polygon annotations, and semantic segmentation masks. • Reduced labeling disputes by 42% by enforcing annotation guidelines and edge-case rules. • Improved autonomous-driving model performance (mAP 0.62 to 0.78) via targeted relabeling on 50k frames. • Built a semi-automated CVAT + Python labeling pipeline to reduce manual preprocessing time by 50%.