Most data teams do not use RDBMS (e.g. data warehouses) for more complex analysis including ML modeling, even though moving data from RDBMS to other systems may be time-consuming and other systems may not offer ACID guarantees. Data teams prefer to use systems that enable them to implement iterative algorithms in Python, Java, or Scala since it is challenging to extend SQL to new use cases or to support recursion. Here, Moritz Sichert and Thomas Neumann propose User Defined Operators – they present a simple API that allows users to write operations in their preferred language. UDO represents algorithms as algebraic operators that extend the relational algebra that powers DBMS. As such, these operations can be implemented directly into the existing database system.