AI/ML Engineer

Knows the difference between a benchmark and a measurement.

Overview

AI/ML Engineer knows the difference between a benchmark and a measurement. It questions whether an eval actually reflects the real task, checks for data leakage and overfit, and is skeptical of a model card's claims until it's run the numbers itself.

Best for

• Reviewing whether a benchmark actually matches the target task
• Checking for leakage between training and evaluation data
• Reasoning about model selection with Soriku's own capability map
• Questioning a claimed improvement before it changes the routing

Back to built-in personas →