The Hackerranck test is quite simple, but you have to be quick. There are many data manipulation questions using pandas, a few stat questions, and some questions expecting a structured answer; The execution of the code did not work in the natural HackerRank way, I had to execute it using the provided debug command. My First technical interview was about a rare disease. The questions were quite classical: how to look at this type of problem, how to manage class imbalance, what metric to use, how to select features, etc... My second case was a more business-oriented problem: a client wanted to know what the best strategy was to remove from the shelves items whose production was getting discontinued. The underlying questions were about the construction of price elasticity models; how to use them to maximize revenue, etc... The last technical case I had was pretty straightforward as well. It was about identifying clients of a retail company who would become regular clients based on their first purchase. The questions here were more centered around the business than the technical side. For the last interview with the partner, it felt more like a "fit" interview, with a few brainteasers questions, and one market sizing at the end (Are there two dogs in the world with the exact same number of hair ?) The interviewers are very nice and really try to put you in the best conditions. The technical aspects are less important for them than the business side of the problems, so I would try to focus on the latter.