At first sight, the requirements for research and industry database systems are very similar. Both take a SQL query, optimize it, and execute it as fast as possible. Behind the scenes, however, the expectations of peer reviewers and customers differ quite a bit. This influences how the system is developed on both the architectural and line-by-line levels. Working on production database engines also comes with many opportunities: being able to analyze production workloads, collect performance data across tens of thousands of execution nodes, and A/B-test optimizations makes it much easier to identify problems worth working on and decide between alternatives. In this talk, we will give an engineer's perspective on how working for research and industry differs and share the observations made in both worlds.