CapitalOne Onsite Interview Experience for Applied Scientist Position

There were 6 rounds in total. The following content requires a score of 166 or higher. You can already view it.

Round 1 LLM-related standard questions, not many.

Round 2 Talk. Three more senior inte

There were 6 rounds in total. The following content requires a score of 166 or higher. You can already view it.

Round 1 LLM-related standard questions, not many.

Round 2 Talk. Three more senior interviewers interrupted to ask questions. I heard some case studies were asked repeatedly. They asked some questions during my talk, but not as many as I expected. Rounds 3 & 4: Behavioral questions. Three questions per person. Collect enough signals.

Round 5 Case study. Understand the behavior of a piece of code in a hypothetical scenario, and modify the code according to updated requirements, or find bugs and improve the code. As long as you are careful and patient, you should be fine.

Round 6 ML knowledge. I thought this would be a standard ML discussion, but it turned out to be an ML + behavioral interview. They asked me to give some examples from my previous work to demonstrate the complexity and scope of my work, and asked some practical questions about how to handle scenarios, such as how to monitor distribution shifts in online systems, when to retrain, and how to plan debugging when encountering certain problems. Overall, I think the ML knowledge and talk rounds were the most crucial. The interviewers in the other rounds didn't mean to make things difficult for me. It felt like we were just forming a form together. The team is working on a task. Finding a job has been tough, so please give us some points!