Capital One Intern Data Scientist Interview Experience — CodeSignal, HM Call, and a Four-Round Powerday

1. Application: Applied March 5th, recruiter replied on the 11th asking for details to confirm qualifications, and I got a DSA (Data Science Assessment via CodeSignal) immediately, to be completed within 2 weeks.

2. Assessment: 4 questions overall, 90 mins. 3 questions of DataFrame manipulation using pandas on a real-world tabular dataset, fairly straightforward but they can become medium difficulty under time constraint. For the pandas questions, practice on StrataScratch. The last question was open-ended: implement a model meeting a desired performance threshold. Practice writing a simple pipeline: load dataset, train/test split, train model, evaluate. Scored in the competitive range. The OA can vary, these are the questions I encountered but be prepared for other variants.

3. HM Screening: Roughly 8 days after completing the OA, I connected with the HM for a 30-min phone call, no Zoom. We discussed my experience and recent projects and how they work. The HM may discuss the team as well.

4. Powerday Overview: I got a mail for Powerday scheduling shortly after the HM screening. Four rounds: Case Analyst, Technical, Stats Role Play, Job Fit. The Powerday guide the recruiter shares is very helpful. Pro tip: you'll be given an option to schedule a prep call with a recruiter before Powerday, use it to understand what skills to focus on for each round. They are very helpful in setting expectations.

5. Actual Powerday:

Case Analyst: The Powerday guide and its examples are all you need. Communicate every step to the interviewer, they are open to correcting you if you make mistakes. Note that information is given verbally, not visually, so practice the guide examples by having someone read them to you.

Technical: Focused on code review and testing scripts. Know what a .sh file is and how it works. The functions you review are abstractions of sklearn-style classes, so review fit/transform patterns and understand why class-based design is preferred in DS settings. Know about testing and how to write proper test coverage.

Stats Role Play: The interviewer presents a use case and gives you time to review data before walking through your findings. Learn A/B testing, statistical significance, and p-values, StatQuest videos are excellent for this. Ask clarifying questions about metrics for use cases. Prepare to analyze regression tables. You are assumed to be explaining to someone with no statistical background, so practice explaining p-values and R-squared in plain language.

Job Fit: Generally you're expected to get STAR-format experience questions, but this can vary significantly by candidate and interviewer. In my case it became a technical discussion covering LLM architecture, agent design, and AI ethics. Utilize the prep resources your recruiter provides and ask what to expect, this round can look very different depending on who interviews you.

Overall: Interviews are back to back and demanding. The schedule is provided the day before. Technical is not LeetCode style, your recruiter will explicitly confirm this. The Ace the Data Science Interview book's ML section is good prep. The best resource is your recruiter prep call, use it.