Prompt + Response Evaluation & Enhancement
The project focused on evaluating and improving AI generated responses for a customer support chatbot using a single prompt + response writing format. Reviewed and rated responses for accuracy, relevance, helpfulness, and tone, rewriting them when needed to create a high quality dataset. Processed around 5000 prompt response pairs monthly, following strict quality measures including clear guidelines, inter rater checks, spot reviews, and automated QA.