I was contacted by a recruiter to apply for a data science position. Initial phase was a coding challenge. They sent a dataset, and you were expected to produce some results and answer some questions. Pretty typical. Next phase was a phone interview with one of the data scientists. Basically walked through one sampling/probability type problem for about 20 minutes and then answered some data science systems level type problems.
After that, they flew me out to have an onsite interview. The place they put you up in overnight is really nice. The interview lasted about 5 hours in total, lunch included. This consisted of several one-on-one interviews with a handful of the data scientists and a data science manger. Primarily systems level, experimental design type questions, with some technical deep dives when necessary. To be honest, I think the conversations might have gotten more technical if I had done better in my interviews, so YMMV. Was in the wrong frame of mind going in; I was thinking and prepping very technical material, lots of coding, machine learning and the discussions were pretty much on the other side of the spectrum. Little to no coding questions through the whole process, which I was surprised by.
The whole process was great, and the company seems really awesome. I wish I had done better! :)