RWS Anotation
Served as a lead annotator for a high-volume data pipeline focused on improving the performance and safety of large language models. The scope of work involved complex RLHF (Reinforcement Learning from Human Feedback) tasks, including prompt evaluation and response ranking. I performed detailed SFT (Supervised Fine-Tuning) through text generation and multi-turn conversation modeling. Quality was maintained through strict adherence to evolving project guidelines, ensuring all outputs met rigorous standards for factual accuracy, linguistic fluidity, and safety alignment.