diff --git a/_questions/data-engineering-zoomcamp/module-3/015_a198f6959c_layered-data-warehouse-batch-pipeline.md b/_questions/data-engineering-zoomcamp/module-3/015_a198f6959c_layered-data-warehouse-batch-pipeline.md
new file mode 100644
index 0000000..edd5daf
--- /dev/null
+++ b/_questions/data-engineering-zoomcamp/module-3/015_a198f6959c_layered-data-warehouse-batch-pipeline.md
@@ -0,0 +1,18 @@
+---
+id: a198f6959c
+question: How do I structure a layered data warehouse (raw, clean, analytics) for
+  a batch pipeline?
+sort_order: 15
+---
+
+- Raw layer: store ingested data exactly as received
+- Clean layer: filter invalid records and enforce basic constraints
+- Data quality (dq) layer: validate completeness and consistency (e.g. missing timestamps)
+- Analytics layer: build aggregated views
+- Mart layer: expose final business metrics
+
+Flow: raw → clean → dq → analytics → mart
+
+Each layer is implemented as SQL transformations in PostgreSQL and BigQuery.
+
+This separation helps with debugging, testing, and ensuring that analytical outputs are built on validated data.
\ No newline at end of file