AI Chatbot Response Quality Annotation
Annotated and rated AI-generated responses for accuracy, helpfulness, tone, and safety across a range of query types. Compared multiple model outputs and selected or ranked the best responses based on defined preference guidelines. Provided granular feedback labels to support RLHF (reinforcement learning from human feedback) pipelines, completing tasks with high consistency scores throughout the project.