AI / UX Evaluation Work (DataAnnotation / Outlier / UserTesting) — score and evaluate AI outputs and usability
Scored AI-generated responses for factuality and reasoning quality against structured rubrics while identifying common loss patterns that explain how and why model outputs fail. Evaluated model outputs for accuracy and preference in visual and audial comparison tasks. Conducted structured usability evaluations of digital products to surface UX failures and friction points with actionable written feedback. • Factuality and reasoning quality scoring using rubric-based evaluation • Loss pattern analysis to document failure modes • Preference/accuracy judgments for comparison tasks • Usability evaluation write-ups for UX findings