Data Labeller / Content Evaluator — Mindrift AI (Feb 2025 – Present)
Labeled AI-generated content for model fine-tuning and evaluation across diverse subject domains. Performed multi-label classification, binary preference tasks, and open-ended response rating at scale. Identified and flagged hallucinations, bias, and harmful outputs and collaborated with QA reviewers to resolve discrepancies and maintain inter-annotator consistency. • Multi-label classification • Binary preference labeling and response rating • Hallucination, bias, and harmful-output flagging • QA discrepancy resolution and inter-annotator consistency