Data Quality Reviewer (Egocentric Data) — Atlas Capture Program
Served as a Data Quality Reviewer for an egocentric data annotation pipeline, auditing high-volume video and associated textual labels against strict structural and behavioral guidelines. Evaluated and scored initial annotations to catch subtle language, contextual, and taxonomy/framework issues before data is used downstream. Provided clear written feedback describing discrepancies and corrective actions to improve dataset quality. • Reviewed egocentric video and text for 100% alignment with annotation guidelines. • Evaluated physical interaction descriptions and verified label consistency. • Maintained reviewer accuracy benchmarks and quality quizzes. • Authored structured feedback to optimize downstream training sets.