GreenStars annotated 1,500 audio recordings and corresponding English and Kinyarwanda transcripts to create a structured, machine-readable dataset for training an AI-powered chatbot.

The assignment covered intent classification, entity tagging, and speaker identification, supported by standardized guidelines, bilingual annotation, continuous quality control, and independent quality assurance. Key deliverables included the fully annotated dataset, annotation guidelines, and a quality assurance report.