I think it is healthy to think of DataRobot as an additional tool in the datascience toolbox which helps with convenience. While DataRobot has a lot of great features which are usually cumbersome to do manually, it does benefit to have datascience knowledge when using the software. Knowledge about how many CVs to do, how large a holdout is important, along with the general working of the different models and types of regularization is important.
Another thing to note is the Python API which I am loving. Training a bunch of models on the cloud with different splits with only a few lines of code is massively more convenient than loading all the various packages and doing all the model setup manually.
Interpretation of ROC/Lift curves, residuals and other metrics is also very important when scrutinizing the different models.
All in all, my experience tells that you don't have to worry about not using your data science skills.