I received a Data Science Challenge within a week of applying. The challenge was an half-an-hour test on HackerRank. The challenge tested conceptual knowledge of machine learning algorithms and statistics. Common topics of testing were clustering, overfitting and hypothesis testing.

Start with a phone call with HR. They ask generic questions on past experience, had the feeling they tried to match keywords against a checklist. They really want you to name some ML models. The second step was an online multiple choice test on Hackerank. 12 questions, about 80% about statistics, 20% on basic machine learning (cross-validation, feature engineering). The third step was a 45-60 min skype call with 2 Data Scientists. The internet connection was unstable so we backed up on a regular phone call which didn't allow for proper communication. The initial questions are about your experience with ML models: they test your knowledge on as many details as possible. The second part of the interview is solving a business case on the fly. You explain how you'd approach it and why.