AI Product Engineer, 51Talk — AI Efficiency Center
Served as the lead for quality control reviews of image annotations, systematically identifying and correcting inaccurate or missing labels in large-scale computer vision datasets. Developed and maintained a Python-based data classification pipeline to improve annotation precision and align data outputs to business requirements. Authored annotation quality guidelines and review checklists to standardize labeling workflows across teams. • Led detailed reviews of auto-generated image labels, ensuring strict accuracy and completeness. • Designed AI training programs and annotation best practices, adopted by over 500 learners. • Configured annotation quality standards and iterative validation processes with cross-functional input. • Used CVAT and Label Studio extensively for annotation tasks and workflow management.