E.g. designing scientific instruments. The fundamental physics and chemistry can be well understood, and yet you need a strong overlap of scientists and engineers to produce and run something that actually collects useful data, especially at the cutting edge, where new things actively need to be discovered and built to achieve the desired capability. Another growing one is using AI to drive scientific discovery (e.g. sifting through the terabytes of data being generated everyday and identifying things of potential interest), it isn't strictly an engineering problem, as the entire point is that you don't fully know what you are/are not looking for.
There's a reason most of the things I mentioned also hire plenty of physicists.
If a team of engineers find a cool new algorithm to make computer vision easier, we learnt something new about the world in the process. On the flip-side, you actually have plenty of research in fields you would consider science, eg. physics, that do not use the scientific method at all, but instead deduce possibilities based on mathematical modelling.
