Yazaki's multi-source supply chain lakehouse ingests demand forecasts, stock levels, pricing, and bill-of-materials data via Databricks Auto Loader into a medallion architecture spanning Bronze, Silver, and Gold Delta Lake layers. Raw CSV and Excel files from SAP exports and manual deposits flow through Auto Loader's cloudFiles and schema management into Bronze tables, then transform through Silver cleaning and aggregation into Gold analytical tables powering KPIs like stock coverage, shortage indicators, and valuation metrics. Machine learning models built with PySpark and Spark ML predict shortages using features extracted from the lakehouse, with MLflow tracking experiments and model performance across accuracy, precision, recall, and AUC-ROC metrics. Unity Catalog enforces governance, lineage tracking, and data quality across all layers, while Lakeflow orchestrates Bronze-to-Silver-to-Gold-to-ML job dependencies. Power BI consumes the Gold layer through a semantic model with fact and dimension tables, delivering supply chain visibility, shortage detection, and financial analysis across five interactive pages. Fork this diagram on Diagrams.so to customize data sources, add real-time streaming, or extend ML features for your automotive supply chain.