Whenever you're ready, walk me through how you'd approach this.
Tell me about how you would resolve a situation where a Spark job takes too long to execute, and how you would use Spark UI to diagnose it.
Explain a practical, systematic investigation from job symptoms through remediation. Cover how you distinguish input, CPU, memory, shuffle, skew, serialization, and cluster-resource problems, and how you validate that an optimization improved runtime without changing results.