Genie Benchmark Evaluator

Evaluates Genie Space responses using a multi-dimensional 3-layer judge architecture with MLflow tracking. Supports both Databricks Job and inline evaluation modes.

When to Use This Skill

• Scoring Genie Space accuracy against benchmark questions
• Comparing evaluation results across optimization iterations
• Running post-deploy verification after bundle deployment
• Testing repeatability of Genie SQL generation

Inputs (from Orchestrator)