Applied on State Farm website, received automated email with exercise instructions and a link to data sets. Assignment was straight forward: build 2 classification models to predict a binary outcome and submit a brief write-up on the differences between the 2 methods. Models are scored using AUC. Around 100 mostly numeric and some categorical features. Nearly all fields had missing data which require remediation. One model I built showed a nearly perfect decision boundary (AUC = .99) when tested against hold out data. Assignment was prescribed on Friday and due on Monday. Code was very clean, used a novel imputation technique, achieved extremely high validation scores. Got an automated email 5 days later (Friday) saying I did not achieve a competitive score. I was highly skeptical given my experience and careful validation before submitting the assignment. I asked for a more detailed explanation and received feedback saying the code was strong and one of the models scored above their threshold (which I knew it would). The feedback indicated they were looking for certain things in the data exploration and analysis although I can’t imagine what it might be given the large amount of seemingly nonsensical features, lack of context, and limited amount of time. Not sure what the game is but they are not looking to hire data scientists. I've heard that the hiring manager over there is a bit of a tool and is impossible to work for...I can see why. My suggestion is to not waste your time, there are PLENTY of other opportunities for skilled data scientists that won’t waste your time.