Senior RLHF & Large Language Model Training Specialist
Text
RLHFLed large-scale RLHF and AI training initiatives supporting the development of advanced large language models across multiple domains including education, business, technology, science, and creative writing. Evaluated and ranked AI-generated responses based on accuracy, factuality, reasoning quality, instruction adherence, safety, and user usefulness. Annotated and reviewed over 1.2 million text samples, prompts, and model outputs using detailed annotation guidelines and quality assurance frameworks. Conducted pairwise ranking, preference modeling, hallucination detection, factual verification, and response quality assessments to improve model alignment and performance. Collaborated with cross-functional teams to refine annotation taxonomies, improve reviewer consistency, and develop quality control procedures that increased inter-annotator agreement rates to over 97%. Regularly performed dataset audits, root-cause analyses, and error categorization to identify weaknesses in model behavior and improve training data quality. Contributed to multiple model releases by generating high-quality human feedback data used for supervised fine-tuning (SFT), reinforcement learning workflows, and safety evaluations. Maintained an average quality score above 98% while consistently exceeding productivity targets across high-volume annotation projects.