Yes! AQE helps, but it does not remove the need to read physical plans. In Databricks, the practical skill is to use the plan to spot data movement, skew, sort-merge joins, and unnecessary shuffles, then decide whether to change partitioning, file layout, or join strategy; AQE and Photon can help automatically, but they still work within the shape of the query you give them.

Say in interviews
“I use the physical plan to find bottlenecks, then tune the data layout and query shape so Spark can execute efficiently at TB/PB scale.” AQE re-optimizes during execution based on runtime stats, but it cannot fully fix bad table design, extreme skew, or a query that reads too much data in the first place.

Short resources
For practical reading, start with Databricks AQE docs, Spark SQL tuning docs, and Databricks docs on Photon and Unity Catalog pipelines; those cover the exact runtime behavior you’d discuss in interviews.