Whenever you're ready, walk me through how you'd approach this.
What are some best practices for data modeling?
Explain how you would design models for an ETL pipeline supporting analytical workloads. Cover grain, keys, normalization versus denormalization, naming conventions, historical changes, partitioning, data quality, lineage, and performance. Include a simple example showing how raw source data becomes a well-structured dimensional model, and mention how you would validate the model in production.