Lucas Martins

I evaluate AI-generated responses in Brazilian Portuguese and English, assessing correctness, relevance, completeness, clarity, factuality, and instruction following.

Through independent bilingual AI evaluation practice, I've compared model outputs using rubrics, identified unsupported claims and hallucination-like statements, classified error severity, and written evidence-based rationales. I also use research and fact-checking workflows to verify claims before making evaluation judgments.

My background in administrative, billing, logistics, and operational work strengthens my business-domain reasoning and attention to structured information. I've also planned, drafted, revised, and finalized a 287-page original novel, bringing sustained writing and editing discipline to AI training work.