First round was more behavioral in nature, really focusing in on what motivates me and the issues I care about and would be interested in working with. I got to know the people there in a way that seemed very genuine. Second round got a bit more technical, but still mostly conceptual. For example, I was asked how I would deal with very large feature spaces without adequate row count. I explained, verbally, some potential approaches (PCA, unsupervised feature reduction) and was pushed on detailed questions about how the processes work but was not required to provide mathematical proof or code. To me this was a perfect balance.