AI Trainer, Alignerr (Anthropic)
As an AI Trainer for Alignerr, I contributed to the development of AI models using the Python programming language. My role focused on providing human preference feedback and engaging in reinforcement learning from human feedback (RLHF) tasks. This work aimed to improve the effectiveness and safety of Anthropic's language models. • Performed pairwise comparison and rating of AI-generated code outputs. • Used internal and proprietary annotation platforms designed for AI training tasks. • Helped refine complex prompts and responses. • Collaborated with domain experts to optimize labeling processes.