QueryGuard

QueryGuard is a web app (with optional CLI) that monitors and enforces “safe query” policies across modern data warehouses. It detects expensive or risky SQL patterns (missing partitions, cross joins, unbounded scans, bad clustering usage), predicts cost and runtime before execution, and can block, throttle, or require approval based on rules. It integrates with Snowflake, BigQuery, and Databricks SQL via audit logs and query history, then pushes alerts to Slack/Teams and tickets to Jira. The MVP focuses on visibility and guardrails: per-team budgets, top offenders, and a lightweight review workflow for high-cost queries. This is a combination traditional + AI app: traditional rules for deterministic enforcement, plus an AI layer that explains why a query is risky and suggests safer rewrites. It’s realistic because it starts with read-only monitoring and incremental enforcement.