Key Points

• Flink is used when you need continuous computation over streams rather than batch processing after the fact.
• It is strong for windowed aggregations, fraud detection, anomaly detection, and other cases where timing and state both matter.
• Event time, watermarks, and stateful operators are the ideas worth knowing at interview level.
• Flink usually sits downstream of a broker like Kafka and produces derived streams or materialized outputs.