AI Data Trainer & Technical Annotator (Remote) — AMME Company
Provided structured data labeling workflows for complex multimodal datasets, including text classification, entity extraction, and multi-turn dialogue scoring using dedicated annotation tools. Audited AI-generated code and structured outputs for syntax correctness, logical gaps, and edge-case constraints, documenting failure patterns for model improvement. Applied multi-step scoring criteria to evaluate and rank LLM reasoning chains and reduce downstream reasoning/code-generation errors. • Labeled and scored entities and multi-turn dialogue data for evaluation datasets • Reviewed Python/SQL/C outputs for logical correctness and constraint adherence • Cleaned datasets and reduced label noise using automated Python tooling • Managed high-volume bulk annotation runs to maintain labeling reliability