One reason for learning APL mentioned in the interview is learning to solve problems without branching. Conor's knowledge of solving problems in a data paralle

One reason for learning APL mentioned in the interview is learning to solve problems without branching. Conor's knowledge of solving problems in a data parallel fashion using masks and such in APL helped him with his day job of building GPU accelerated data-science tools at NVidia. But it's not just a NVidia thing - CPUs can do data parallel operations and also get slowed down by branching. I don't think this means that you should write everything in an array language, but it means that learning how to solve problems under those constraints will teach you important lessons.