AI Expert Trainer at Outlier AI (Remote)
Text
Fine-tuningTrained and refined large language models by evaluating, fine-tuning, and improving model outputs for accuracy, coherence, and safety. Wrote and assessed complex prompts to sample and test model abilities across tasks including creative writing, technical reasoning, and factual questioning. Performed quality assurance by rating responses against predefined rules covering helpfulness, truthfulness, and potential bias. • LLM fine-tuning and model evaluation for quality and safety improvements. • Prompt engineering with written prompt creation and response review. • Response quality evaluation using structured rubrics for helpfulness, truthfulness, and bias. • Collaborated with distributed AI trainers and researchers to identify weaknesses and improve performance.