Voice AI Trainer
Participated in a Voice AI evaluation and data labeling project focused on comparing real-time conversational models in Vietnamese. The work involved recording and analyzing multi-turn conversations across various everyday scenarios (health, lifestyle, history, travel, and consumer decisions), evaluating goal completeness, task efficiency, conversational smoothness, pronunciation accuracy, and overall user satisfaction. Each task required structured qualitative feedback, evidence-based comparisons, and adherence to strict QA guidelines. Quality was ensured by following standardized evaluation rubrics, providing clear explanations for all ratings, and aligning feedback with linguistic, contextual, and usability criteria to support model improvement.