AI Trainer & Data Evaluator – LLM Alignment
Text
RLHFCollaborated on a high-priority AI training project focused on fine-tuning and aligning next-generation Large Language Models (LLMs) for safety, accuracy, and helpfulness. Key Responsibilities & Tasks: • Evaluated and ranked AI-generated responses based on complex multi-turn prompt guidelines, focusing on truthfulness, harmlessness, and nuance. • Performed rigorous RLHF (Reinforcement Learning from Human Feedback) tasks, providing detailed written justifications for preference rankings. • Annotated and labeled diverse text datasets for specific domain knowledge, including logical reasoning, creative writing, and factual consistency. • Conducted adversarial prompting (red-teaming) to identify model vulnerabilities, biases, and edge cases. Quality & Impact: • Maintained a 95%+ quality accuracy score across peer reviews and internal audits. • Strictly adhered to complex, evolving taxonomy guidelines under tight deadlines.