There is a 3-level interview process. 1) A brief discussion with the project manager. You will go through your CV and delve deeper on projects that appear interesting. No technical questions at this stage, only a few according to what you have done in your past/current positions. 2) Choose any topic of Machine Learning and prepare a ~15' presentation. This stage takes 1:30 hours. You present your topic of choice and then they ask you questions on your presentation. For the last 30' you will have a brief discussion with people from other departments who are there to assess your interest / passion about the role. Note: During the presentation you must delve deeper and deeper on a project you did, show code and results. Do not spend time only on different approaches or do not perform an introduction to X topic of ML. They want to address your level of experience. 3) Technical Task (NLP). Simple. Try to be precise as much as possible. Not so much EDA. Instead keep it short and clear. Overall it was a really nice process and the people were very open to see your point of view and they respect you the whole time. The only negative aspect is the absence of details on what they want exactly. Specifically, the presentation and task are not as clear as you would assume/hope. Especially the presentation part gives a large advantage to people who have been doing research and they can thus, present their problem and process. In my case, it was not clear from the start that they needed code and especially, presenting a project. Additionally, due to confidentiality agreement I could not reveal much of my work. During the feedback, they told me that this point lead in rejecting me, as I did not show what I can do so they assumed that I do not know much.