Apache Spark
Unified compute engine for big data — Matei Zaharia (UC Berkeley AMPLab), 2009, donated to ASF in 2013. Much faster than MapReduce thanks to in-memory caching and DAG execution.
Four parts: Spark SQL (DataFrames + Catalyst optimizer), Structured Streaming (micro-batch streams), MLlib (machine learning), GraphX (graphs). Written in Scala, first-class APIs in Python (PySpark), R, Java.
Commercialised by Databricks. Competitors: Trino (interactive SQL), Flink (stream-first), Snowflake (managed SQL warehouse), DuckDB (single-node). Spark still dominates "industrial big data".