I applied online. The process took 7 weeks.

My process consisted of 6 interviews and lasted around 7 weeks, though I think there are normally 5 interviews. All of them were in English, except when speaking with the recruiter, which was in Spanish. All technical rounds were done live sharing my screen with HackerRank.

• Recruiter Phone Screen: A couple of days after submitting my CV, I was scheduled a 30-minute interview with the recruiter to talk about the process, the position, and my experience. Here, I think he decided to schedule a preliminary interview with the hiring manager (actually, 2 HMs showed up in the interview) to assess experience, which happened around 3 weeks later.
• Hiring Manager Interview 1: We talked about my previous projects (both on software engineering and data science), experience, and expectations, and I was asked to briefly describe (5 minutes or so) how machine learning can be used to help route customer support tickets. On the next day, I was sent a PDF with the description of the process and was scheduled for the next 2 interviews almost 3 weeks later.
• Basic Programming Interview: It consisted of a set of 4 LeetCode-type problems to be solved on HackerRank while sharing my screen. The first 3 were really simple and were more to warm up, and the last one was a bit more complicated, but still LeetCode easy level. The interviewer was a really nice guy and made the pressure go away. He gave me the option to speak aloud my process (which I did) or to solve it all by myself and then share my comments. He told me that if I had struggled, he would have guided me to reach a solution.
• Architecture Case: I was given a business problem, and we discussed how to pose it as an ML problem and about the microservices we would need to reach a good design. I don't know if it was planned, but it was very similar to the customer support question I was asked on my second interview, but this time the question was on how to be more proactive and how to prevent questions from reaching human interaction. The interviewer was also very nice and he was very open to communication.
• Hiring Manager Interview 2: I spoke again with one of the hiring managers of the first interview, and in fact it was very similar, but went more in depth with my previous projects, my experience, my motivations to join, and technology.
• MLE Programming Case: I was given a fully working code that loaded a trained model from a Pickle and produced an output, and I was asked some questions on how to use that model at a business level and to make some changes to implement new features. It also had the option to implement unit tests, but the interviewer told me they were not required and we could only discuss them and that it didn't matter if we didn't manage to solve the 4 questions in time (but we covered like 90% of it). This was like pair programming with no ML theory (but a bit of statistics) and only to assess how you can use an existing ML model. It was the same interviewer as in the basic programming and he was really cooperative and gave me hints when I needed them. He also helped me to debug one of the questions.
• Behavioral Interview: This was with the recruiter and was about my previous and current projects, things I have learned, and how I have handled different situations related to soft skills. In this part, I felt that interaction directly with stakeholders is highly valued. He also asked my motivations to join and told me that all the interviewers would gather to debrief in the next couple of days and decide if I was a good fit for the team and which level of seniority they could offer.

You were given an array as an input and were asked to find the minimum value of an integer X such that the running sum (X = X + array[i] for every element of the array) is always >= 1.

Why Nubank? Are you a Nubank client? What are the pros and cons of working with you? How do you handle constructive feedback?