AI Training & Annotation Specialist (Freelance / Tutor Context) — self-employed
TextProvided guideline-based evaluation of student legal arguments and text responses using strict structural rubrics aligned to AI labeling and schema verification. Delivered multi-tiered response critique with factual corrections and syntax enhancements across hundreds of long-form academic and legal texts to simulate preference metrics and RLHF-style assessment. Conducted hallucination and error detection by reviewing legal documentation and citations to isolate false claims and factual inconsistencies for accurate model evaluation. • Evaluated responses against complex rulebooks and explicit taxonomies. • Performed quality checks including fact-checking and policy/safety violation vetting. • Applied RLHF concepts including response preference ranking and alignment criteria. • Managed large batches of training content and learner records to meet quality targets.