Bronze-to-Silver Cleaning Agent

This agent cleans raw bronze layer data by applying null filtering and type corrections, then stores the cleaned data in the silver layer. It analyzes the raw data schema and values to infer column-level cleaning logic such as null removal and type coercion, and makes SQL decisions based on null patterns and schema inferences. Once the SQL is generated, it executes the transformations and stores outputs without human involvement.

The agent addresses unreliable and inconsistent raw data by automating data sanitation, so downstream processes rely on trustworthy data. It reduces manual cleaning effort, improves data quality, and accelerates pipeline readiness by 60%. Its primary users are data engineers and data quality management teams working with large datasets. It integrates with BigQuery for processing and Cloud Storage for temporary artifacts, and uses the Vertex AI Gemini model for SQL generation.