Foundations of distributed data

Why this track exists: Spark, Kafka and Flink are three answers to the same few problems: partitioning, replication, ordering, time and failure. If you learn those problems once, each tool becomes a set of design choices instead of an API to memorize.

Lab profile: none (pure Python) plus core for FND-07.

Interview drills: "How would you design for exactly-once?", "Explain watermarks to a PM", "Why columnar?"