A data engineering team is spreading data across a Hadoop-based data lake and a warehouse-powered BI platform, and wants analysts to query all of it consistently without dealing with the underlying system complexity. Which architectural approach directly addresses this goal?