Capital One Interview Questions — Page 9

Capital One·Data Analyst·Technical Phone Screen

May 2026

Capital One Data Analyst case interview built around a restaurant's decision to partner with Groupon. It was mostly unit economics with a side of business judgment, and the interviewer kept pushing on whether I was reasoning about net contribution or just top-line traffic.

Before running any numbers, what factors should a restaurant think through before deciding to partner with a daily-deals platform like Groupon?
What is the restaurant's current daily operating profit given 20 tables, $30 average spend, 40% variable cost, and $100 fixed cost per day?
For a purely incremental Groupon customer, what minimum average spend is required for the restaurant to break even on that table?
Would you recommend the restaurant partner with Groupon? Explain the difference between an incremental customer and a cannibalized full-price customer, and make the comparison quantitative.
Now suppose the restaurant serves 25 tables at $36 average spend, with 10 of those being Groupon tables. What is the new daily profit, and why might it be lower than the original $260 despite higher table count and higher average spend?
Under what conditions would you still pursue a Groupon partnership despite the short-term profit hit, and what specific levers could the restaurant use to improve the unit economics of the deal?
If you could run an experiment to measure the true incremental value of a Groupon partnership, how would you design it?
How would you fold repeat customer value into the per-table economics if, say, 30% of Groupon redeemers return at full price within 90 days?

“I jumped straight to margin math and the interviewer had to pull me back.”

View Post

Capital One·Software Engineer·Technical Phone Screen

May 2026

Capital One SWE interview with a tricky array rotation problem. Not the hardest thing I've ever seen but the edge cases were sneaky and I didn't feel great walking out.

Given an array and an integer t (where t is less than the array length), a cyclic left-shift moves t elements from the end to the front. Can you determine whether, within t such operations, the array reaches a descending order at any point?

“Spent the first few minutes just re-reading the problem because the shift direction tripped me up.”

View Post

Capital One·Machine Learning Engineer·Technical Phone Screen

May 2026

Capital One ML engineer interview with a practical ML design question. Nothing too wild but it required more depth than I expected going in.

You have a dataset of historical deliveries. How would you build a model to predict delivery times?

“I went straight to regression and started listing features like distance, time of day, weather.”

View Post

Capital One·Software Engineer·Technical Phone Screen

May 2026

Capital One data engineer interview, one SQL round that was heavier than I expected. The question looked straightforward on the surface but the join-plus-window-function combo with tie-handling tripped me up a bit.

Given an accounts table and a transactions table, find the top 3 customers per segment by total transaction amount over the last 30 days. A customer can have multiple accounts. Use window functions to rank within each segment, and include all tied customers at rank 3.

“I got the basic join and the GROUP BY done pretty fast, but then sat there for an awkward moment on the tie-handling part.”

View Post

Capital One·Data Scientist·Onsite - System Design / Architecture

May 2026

Capital One data scientist loop, and this one was a beast. The whole interview was essentially one giant case study about a broken registration pipeline and they wanted a full plan from metrics to architecture to postmortem process. Walked out not totally sure if I nailed it or completely fumbled the scope.

A premium registration pipeline silently failed for three months, causing thousands of missed registrations and significant revenue loss. As the analytics lead working with engineering, propose a concrete prevention and mitigation plan covering: SLIs/SLOs with alert thresholds and data contracts, architecture changes for reliability, dashboard and runbook design, quantified impact reduction, and a 30/60/90-day rollout plan.

“This is one of those questions where the scope is so wide you can freeze up trying to figure out where to start.”

View Post

Capital One·Data Scientist·Technical Phone Screen

May 2026

Technical screen at Capital One for a principal-level data science role. The whole thing was a code review session where they handed me a shell script and some Python transformer classes and asked me to tear them apart. More opinionated than I expected for a screening round.

Walk through the provided virtual-environment shell script line by line and explain what it does.
What advantages does shell scripting provide specifically in data science engineering workflows?
Given the OutlierHandler class provided, describe its overall purpose.
Why is it useful to separate fit() and transform() into distinct methods in a transformer class?
What coding style or design problems do you notice in the OutlierHandler class?
Write one high-impact unit test you would add for the OutlierHandler class.
Summarize the high-level functionality of the three imputation classes shown in the script.
Identify coding style problems in the imputation script, particularly around the use of 'from numpy import *'.

“I do read shell scripts semi-regularly but explaining them out loud, line by line, is a different skill.”

View Post

Capital One·Data Scientist·Technical Phone Screen

May 2026

Capital One data scientist interview with a case-style marketing finance question. The whole thing was basically applied math dressed up as strategy, which I wasn't fully expecting going in.

If a ride-sharing partnership campaign costs $25 million annually, how many incremental cardholders are needed to at least break even?
If the cost structure changes to $40 per new customer per year plus a one-time $11.8 million marketing spend, how many new customers are needed to hit the annual profit target from the earlier part of the case?

“You need the per-customer profit figure from an earlier part of the case to anchor this.”

View Post

Capital One·Data Scientist·Technical Phone Screen

May 2026

Capital One data scientist interview with a pretty involved economics case around network service contracts. The math isn't hard but the structure they want is specific, and I didn't fully anticipate how granular they'd get with the follow-ups.

Given a monthly revenue of $40 (first 3 months free), a $25/month service cost, $35 install cost, and $120 in marketing and overhead per customer, what is the net value of a customer on a 15-month contract? How does that change if the contract extends to 18 months?
If the contract term is 21 months and a 10% churn rate triggers a $100 penalty per churned customer, while marketing costs shift to $20 variable per customer plus $1M fixed overhead, how many customers are needed to break even?

“I started by computing total revenue across the contract and subtracting costs, but I initially forgot to zero out revenue for the first three months.”

View Post

Capital One·Data Analyst·Online Assessment (OA)

May 2026

Capital One data analyst SQL question, one problem, conditional aggregation with a twist. Pretty clean setup but the logic tripped me up more than I expected.

Given a products table and a purchases table, return the minimum purchase price per product category, but only if that category has at least one purchase rated above 4 stars. If no such purchase exists for a category, return 0 instead.

“The conditional part is what got me.”

View Post

Capital One·Software Engineer·Technical Phone Screen

May 2026

Capital One coding round, one problem the whole time. Pretty standard array manipulation question but the follow-up about non-zero colors tripped me up more than I expected.

You have a zero-initialized array of length n and a sequence of update queries, each setting a specific index to a given color. After each update, return the count of adjacent index pairs where both neighbors share the same non-zero color.

“I got the basic structure pretty fast, the part that slowed me down was handling the non-zero constraint.”

View Post