Learning path
Spark Core
A short background page and five modules, in order. Each starts with a short overview you can reread before an interview; deeper pages hang off it.
HDFS, YARN and MapReduce; how a MapReduce job runs; what Spark solved and what lives on in Spark. Optional if you know it.
Driver, executors, jobs/stages/tasks, the memory model. Deep dive and production memory failures.
Where each runs, why DataFrames beat RDDs, Spark vs pandas, and the problems at the boundary.
Lazy evaluation per action, narrow vs wide, jobs → stages → tasks. Stepper, job-counting practice, production pitfalls.
Optimizer phases, pushdown and pruning, reading explain(), join strategies, AQE, UDF cost.
Debug flow, six failure scenarios, shuffle reduction, partition and executor sizing.