Senior AI Data Annotator & Model Trainer
Delivered RLHF labeling by ranking model outputs and writing ideal responses to support LLM training. Identified harmful or inaccurate generations and applied safety guidelines to ensure compliant preference data. Maintained high accuracy on large-scale annotation workflows across multiple active projects. • Ranked responses using rubrics for helpfulness, factuality, safety, and coherence. • Produced preference and response pairs for instruction-following improvements. • Reviewed sensitive content for toxicity, misinformation, and hate speech. • Sustained 97%+ accuracy and contributed top-10% throughput/quality scores.