Apache Spark is one of the major players if not the leader when it comes to distributed computing and processing. Want to use machine learning to build models and uncover fraud, make predictions, estimate future sales or calculate revenue ? Whip out a 200 nodes cluster on Spark and you are good to go. This talk will show you how to get a shell on each one of these nodes! We are talking about systems that, by design, have access to almost every datastore in the company (S3, Cassandra, BigQuery, MySQL, Redshift, etc.). This is game over for most companies. I will also release a tool that will help pentesters pwn Spark clusters, execute code and even bypass authentication (CVE-2020-9480).