Freelance AI Trainer / Data Contributor (Various Global AI Training Platforms)
Provided structured evaluation and ranking of LLM-generated responses for accuracy, helpfulness, alignment, tone, and formatting constraints. Performed RLHF-style assessment by identifying failure modes such as hallucinations, logical fallacies, and bias patterns to guide refinement. Contributed to maintaining golden-standard expected outputs for consistent model behavior. • Rated responses on factual correctness and helpfulness • Checked alignment with instruction and formatting constraints • Flagged hallucinations, biases, and reasoning errors • Used feedback to improve subsequent training iterations