RLHF Specialist & Data Annotator (Contract) | Outlier | Remote
Rated, ranked, and edited AI model responses for accuracy, helpfulness, tone, and safety. Conducted adversarial red-teaming using complex prompts to probe AI safety guardrails and logical consistency in technical answers. Organized and structured large datasets of technical documentation to improve retrieval and readability for model training. • Evaluated and edited responses for accuracy and safety • Performed red teaming with complex adversarial prompts • Ranked outputs and ensured compliance with safety expectations • Structured technical documentation datasets for training readiness