About This Architecture
Automated ETL pipeline with medallion architecture ingests streaming clickstream data via Kinesis, batch CSV files, and structured documents from RDS through parallel stream and batch processors. Raw data lands in Bronze tier object storage as immutable Parquet/JSON, flows through Silver curation for deduplication and cleaning, then aggregates to Gold tier in Redshift ra3.xlplus for business metrics. Feature Store and Query Cache serve pre-computed analytics to BI tools and analysts without manual intervention, eliminating bottlenecks in the serving layer.