Remote AI Trainer & Prompt Engineer (Independent Contractor)
Don't disclose
Text
RLHF
TextRLHFProvided reinforcement learning from human feedback style AI training support by evaluating, ranking, and critiquing model responses against truthfulness, helpfulness, tone, safety, and constraint adherence. Wrote high-quality reference answers from scratch to establish ground truth for machine learning training sets. Conducted adversarial prompt engineering and red-teaming to detect hallucinations, logical fallacies, and bias in evolving models. • Multi-turn and adversarial prompt design for LLM capability testing • Response evaluation and scoring against strict parameters • Ground-truth reference answer creation for training data • Red-teaming exercises for vulnerability and bias detection
2025 - Present