Generalist AI Trainer and RLHF Evaluator
Text
RLHFAs a Generalist AI Trainer, I evaluated and ranked AI-generated responses for factual accuracy, logical consistency, and instruction compliance. I specialized in reinforcement learning from human feedback (RLHF) by serving as the gold standard evaluator for diverse model outputs, applying cultural and domain knowledge in STEM and general knowledge. I performed data annotation, quality assurance, and contributed clear, guideline-compliant feedback for continuous model improvement. • Conducted prompt compliance checks and logical reasoning error detection. • Fact-checked and verified model outputs related to Kenyan and US cultural contexts. • Ranked AI responses and rewrote flawed answers for multiple domains. • Provided feedback to enhance inclusivity, safety, and relevance of AI systems.