#evals
2 related articles
·3 min
The AI Engineer's Role Has Shifted: From Writing Code to Building Agent Factories
At AI Engineer, Arize AI's Aparna Dhinakaran explained how the AI engineer's role is shifting from writing features to building agent loops—covering triggers, sandboxes, memory, and evaluators.
Read more →
·3 min
How to Ensure Quality of AI Features That Can't Be Unit Tested with Evals
AI features can't be unit tested—the same input yields different outputs. Learn how to measure quality systematically with eval sets: real inputs, written definitions of good answers, three-layer scoring with code checks, human review, and model grading, plus non-negotiable launch bars and production feedback loops.
Read more →