Agentic AI Query Optimization

This agent automatically refactors and optimizes BigQuery SQL queries to reduce costs and improve performance. It applies strategies across readability through a modular CTE structure, data access through early filtration, smart JOIN ordering, and column pruning, storage optimization through smart partitioning and clustering, and execution efficiency by minimizing shuffling and intermediate data.

It is a multi-agent system in which a central Coordinator agent plans a workflow for each unoptimized query and delegates tasks to specialized agents: De-monolith, SQL Optimization, Table Optimizer, and Score/Report. These agents reason about the best strategies and use BigQuery and LLMs to execute them. The result is reduced query execution duration and lower total slot time consumed. Its primary users are data engineering, FinOps, and cloud cost optimization teams. It connects with BigQuery, CI/CD pipelines via Cloud Build, Cloud Composer, and Looker Studio.