Nelson O.

I evaluate AI systems based on reasoning quality, decision integrity, and operational reliability, not surface-level output accuracy. I work at the intersection of AI systems, decision automation, and operational workflows. My background is in system design, growth infrastructure, and diagnostic frameworks, work that requires identifying failure modes, ambiguity, and hidden dependencies inside complex workflows. I apply the same discipline to AI evaluation.Rather than labeling data or performing mechanical annotation, I focus on:How models reason under uncertaintyWhether instructions produce consistent, traceable decisionsWhere confidence exceeds evidenceHow outputs behave when inputs vary or constraints shiftI approach evaluation as forensic analysis, not task completion. My goal is to surface risks, edge cases, and logic gaps before systems are deployed at scale.