Data engineers are often held responsible for making sure that data arrive in the right place at the right time. However, without understanding how the data will be modeled and analyzed by data consumers, they may not architect data pipelines in an optimal way. Here, Vin Vashishta describes how understanding data science can help data engineers determine what metadata is required and how accurate data must be to support downstream applications. In addition, he argues that data engineers must be able to design ontologies that model the systems that data scientists must study.