Data Annotator (AI response evaluation) at Mindrift
Performed prompt-response comparisons to assess AI output safety, helpfulness, and factuality. Reviewed annotation submissions and provided constructive feedback to improve consistency and quality. Ensured responses met evaluation criteria for each prompt task. • Compared AI responses to the given prompts • Checked safety, helpfulness, and factuality signals • Audited annotator submissions for quality • Delivered constructive feedback when issues were found