1Offline vs Online EvaluationOffline evaluation tests a system in a controlled setting; online evaluation tests it with real use, real context, and real consequences.Advanced 2LLM-as-Judge & Its PitfallsUsing one language model to grade another can speed evaluation, but only if you understand what it is measuring and what it can quietly miss.Advanced 3Tracing, Metrics & TelemetryA first-principles guide to the signals that tell you what software is doing, why it is failing, and what to improve next.Advanced 4Regression & Drift DetectionHow to tell whether a model is getting worse, whether the world around it has changed, and what to do before users notice.Advanced 5Building an Eval HarnessAn eval harness turns model quality from a feeling into a repeatable engineering practice.Advanced