Capital One Interview Questions — Page 21

Capital One·Software Engineer·Hiring Manager Screen

Jun 2026

Capital One hiring manager round for a software engineer role, structured like a casual project walkthrough but with behavioral questions woven in. Felt conversational at first, then got a little more pointed than I expected.

Walk me through the project you're currently working on in detail.
Tell me about a technical challenge you ran into on a project, and how you resolved it. Specifically, if another team like infrastructure caused a bug or blocker, how did you approach them to fix it? And if you'd been on that other team, how would you have handled it from their side?
Have you ever had to work with someone who was difficult, and how did you handle it?

“I rambled a bit here.”

View Post

Capital One·Data Scientist·Technical Phone Screen

Jun 2026

Capital One data scientist case interview, heavy on business math and scenario analysis. The whole thing was basically a theme park profitability problem with four parts, each building on the last. Felt more like a finance interview than a data science one, which threw me a bit.

A theme park operates on 2,000 acres with 1,000,000 annual entries across three ticket types (single-day at $80, five-day at $300, annual pass at $1,000). Given variable cost of $22 per entry, fixed costs of $20M, and a land-owner fee of 5% of revenue, compute annual revenue, total variable cost, the land-owner fee, and profit.
Under Expansion Option A, the park acquires an adjacent 1,000 acres, scaling entries and ticket sales proportionally to 3,000 acres total, but the land-owner fee rises from 5% to 10%. Fixed costs stay at $20M and variable cost per entry is unchanged. What is the profit under this option?
Under Expansion Option B, you can bid for the same 1,000 acres. If you win (80% probability), the land-owner fee stays at 5% at 3,000 acres. If you lose (20%), you remain at 2,000 acres with the 5% fee. Compute the expected profit of bidding and recommend the best option among no expansion, Option A, and Option B.
For sensitivity analysis: (i) At 3,000 acres, what land-owner fee rate makes Option A's profit equal to the current no-expansion profit? (ii) Holding the fee at 10%, what variable cost per entry at 3,000 acres would make Option A match the current profit? Show your formulas.

“This part was straightforward once I figured out the entry count for annual pass holders.”

View Post

Capital One·Software Engineer·Onsite - Behavioral / Leadership

Jun 2026

Behavioral round at Capital One for a software engineering role. One question, but it had a lot of layers and I felt like I was still answering it when they cut me off.

Tell me about a time when project priorities shifted unexpectedly mid-way through. How did you respond, communicate the change to your team and stakeholders, decide what to stop doing, and keep the team motivated? What came out of it?

“This question is basically four questions jammed into one and I did not pace myself well.”

View Post

Capital One·Software Engineer·Onsite - System Design / Architecture

Jun 2026

Capital One system design round for a software engineering role. The whole session was focused on one question about building a transaction system for bank accounts, which sounds straightforward until you actually try to talk through all the edge cases.

Design a Transaction class for a bank account system. Walk through the class API (deposit, withdraw, getBalance), how transactions get recorded, handling concurrency and atomicity for simultaneous operations on the same account, error cases like insufficient funds, and extending the design to support transfers between accounts.

“I started with the easy stuff, deposit and withdraw returning updated balances, getBalance, all fine.”

View Post

Capital One·Software Engineer·Technical Phone Screen

Jun 2026

Capital One SWE interview with a data structure design problem that looked straightforward but had a subtle duplicate-counting gotcha. The core challenge was around efficiently maintaining two lists and supporting updates while keeping pair counts correct.

You have two lists A and B. Support two operations: update an element in B to a new value, and given a target t, count the number of index pairs (i, j) such that A[i] + B[j] equals t. Duplicates in either list count as distinct pairs.

“My first instinct was a hash set and I immediately lost the duplicate case.”

View Post

Capital One·Data Scientist·Technical Phone Screen

Jun 2026

Capital One data scientist case interview, pretty heavy on content strategy and business math for a streaming platform scenario. The whole thing revolved around one big multi-part case about show renewals and profit modeling. Walked away feeling okay about the framework parts but second-guessed my expected value discussion the whole drive home.

How would you build a framework for deciding whether to renew or cancel an existing show, going beyond just profit? Think about financial performance, user impact, strategic value, and risk.
How would you estimate the incremental business impact of a show rather than just looking at raw view counts? What biases or confounding factors might mislead you?
Calculate the 2-year profit for the new show in both the success and failure scenarios, then compute the expected profit assuming 50/50 odds.
The existing show produces $50M profit over two years with no startup cost. Given that, which show would you actually choose to back, and why? Don't just use expected value.
What levers would you recommend to improve the existing show's profitability? Cover both revenue and cost sides, and explain how you'd measure whether those levers actually worked.

“This is where I spent the most time and also where I probably over-engineered things.”

View Post

Capital One·Data Scientist·Technical Phone Screen

Jun 2026

Capital One data scientist interview that was basically a live code review session. They handed me Python preprocessing code and asked me to tear it apart, explain design decisions, and write a unit test on paper. More software-engineering-flavored than I expected for a DS role.

What does the OutlierHandler class do at a high level, and why is it beneficial to keep fit and transform as separate methods rather than combining them?
Looking at three provided Imputer classes, what coding style issues or import problems can you identify and critique?
Write a critical unit test by hand for the OutlierHandler class. What would you test and how would you structure the assertion?

“I knew the scikit-learn pattern well enough to talk through it.”

View Post

Capital One·Data Scientist·Technical Phone Screen

Jun 2026

Capital One data scientist interview that was basically one long coding problem about imputation pipelines. The whole thing revolved around leakage prevention which sounds straightforward but the requirements kept stacking up until it wasn't.

Given a user activity DataFrame with numeric, categorical, and time-ordered features plus a binary label, implement fit_imputers() and transform_impute() functions that handle all missing value imputation with strict train/validation leakage prevention.
For numeric features, how would you impute age using country-level medians and income using a regression model, without using the target label or any validation data?
How do you implement time-ordered forward-fill for last_purchase_days_ago and session_length within each user, where propagation stops if the gap between consecutive event dates exceeds 14 days?
For categorical columns like device_type and country, how would you impute missing values using per-user mode with a global train mode as a tiebreaker?

“This is the kind of question that sounds like a clean engineering task until you actually read the spec.”

View Post

Capital One·Machine Learning Engineer·Onsite - Behavioral / Leadership

Jun 2026

Behavioral round at Capital One for an ML Engineer role. Just the one question but it had a lot of layers to it, basically a full story they wanted you to unpack.

Tell me about a time you delivered something that your teammates or leadership thought was too risky or unlikely to succeed. What was the goal, why did people doubt it, how did you approach it, and what kept you going when others were skeptical?

“This is one of those questions where you think you have a good story and then halfway through you realize you're underselling the skepticism part.”

View Post

Capital One·Data Scientist·Technical Phone Screen

Jun 2026

Capital One data scientist interview, one big technical question that basically ate the whole session. It was a credit risk modeling deep dive and they wanted you to think through the entire pipeline from offline eval to production monitoring. Pretty intense for a single question.

You're deploying a probability-of-default model for a consumer lender with a 1% base rate and asymmetric misclassification costs ($1,200 for a missed defaulter, $60 for a wrongly declined good applicant). The regulator cares about stability and explainability. Walk through your three top evaluation priorities, then design the full pipeline: offline evaluation with metric choices, threshold selection under a decline-rate constraint, out-of-time backtesting, a live champion-challenger setup with guardrails, and a production monitoring plan including what you'd do if calibration drifts but rank ordering holds.

“This question is enormous and I think I underestimated how much they wanted actual numbers rather than a framework tour.”

View Post