Explore / Skills

Designing and running evaluations of AI systems

Builds a task-representative test set, chooses graders (rules, human, model-graded with checked agreement), runs regression evaluations across versions and reports results with uncertainty.