Capital One Interview Questions — Page 17

Capital One·Data Scientist·Technical Phone Screen

May 2026

Capital One Data Scientist case interview, heavy on unit economics and sensitivity analysis for a restaurant/Groupon scenario. Four parts, all math, no fluff. The kind of thing where you really need to slow down and not rush the algebra.

A restaurant has 20 tables per day with an average spend of $30 per table, variable costs at 40% of revenue, and fixed costs of $100 per day. What is the daily profit?
Using a Groupon voucher model where customers pay $15 for a $30 voucher and Groupon keeps 40% of that $15, what is the break-even pre-discount spend per table for a voucher customer to contribute zero profit?
On a day with 25 tables total (10 using vouchers, 15 regular), where average pre-discount spend across all tables is $36, calculate the total daily profit using the given formulas for regular and coupon tables.
Holding regular tables at $36 average spend and table counts fixed, what minimum average pre-discount spend per coupon table brings total daily profit back up to the baseline profit from part one?

“Straightforward setup: revenue is 20 times 30 which is 600, variable costs are 0.4 times 600 which is 240, subtract fixed costs of 100, so profit is 260.”

View Post

Capital One·Data Scientist·Recruiter / HR Screen

May 2026

Recruiter screen for a Data Scientist role at Capital One. Pretty standard logistics-heavy call but they packed a lot into it, covering comp, visa, location, and a quick pitch all in one shot. Nothing technical yet.

In 90 seconds, summarize what you're currently working on and quantify the impact you've had in the last six months.
Why do you want this specific role and team? Name the two product areas you're most excited to influence and explain why.
Have you interviewed with Capital One before? If so, when, for what role, and what has changed in your profile since then?
What is your target total compensation broken down by base, bonus, and equity, and what is the minimum you would accept? Are you willing to make any trade-offs between components?
What is your current U.S. work authorization status, including visa type and expiration, and will you need sponsorship now or at any point in the future?
What locations are you open to, are you able to relocate by a specific date, and do you have any constraints around onsite versus remote or hybrid work?
If you received an offer today, what would your realistic start date be?

“The time limit is real and they will cut you off.”

View Post

Capital One·Data Scientist·Onsite - Multi Round

May 2026

Capital One Data Scientist role-play round where you get 15 minutes with a slide deck on airline delay prediction and then have to present to a mock Director of Operations. Pretty intense format, lots of ground to cover fast.

What are the problems with the current feature engineering in this airline delay model, and how would you fix them?
Should this be framed as a regression problem or a classification problem, and what evaluation metrics would you use?
How would you validate this model in a way that respects the time-ordered nature of the data?
Pick two charts from the deck, explain what's wrong with them, and describe what you'd replace them with for a non-technical audience.
Give a concrete recommendation that would reduce compensation payouts by 5% without increasing cancellations, and lay out how you'd measure whether it worked within four weeks.

“The deck had date encoded as a raw integer which is a pretty classic mistake.”

View Post

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

May 2026

Capital One system design round for a Data Scientist role, and this one was a beast. The whole session was basically a single massive question about building an on-device face recognition system at scale, and they kept pulling threads on every answer I gave.

Design an on-device face recognition system for mobile access control with 50 million monthly active users and intermittent connectivity. Should the system do verification or identification, and why?
What model family and embedding dimension would you choose for this face recognition system, and what training objective would you use?
How would you evaluate this face recognition model? Describe your protocol including ROC/DET curves and specific threshold targets.
How would you handle liveness detection and anti-spoofing, including dealing with occlusions like masks and glasses?
How would you ensure demographic fairness in the system, including threshold calibration across different user cohorts?
What privacy and security measures would you implement, covering on-device template storage, differential privacy, template protection, and replay attack prevention?
What are your latency, memory, and battery constraints for this system on a mid-tier mobile device, and how would you meet them?
How would you monitor the deployed system for model drift and manage periodic re-enrollment of users?
How would you safely A/B test a face recognition system like this, including shadow mode rollout and guardrails?

“I went with verification pretty quickly since identification at 50M scale on-device is basically impossible without a server.”

View Post

Capital One·Software Engineer·Technical Phone Screen

May 2026

Capital One SWE interview with a single coding problem that looked like a clean simulation at first glance but had enough edge cases to keep me second-guessing myself the whole time.

You're given an array of warehouse capacities and a list of operations. Each operation is either 'PACKAGE' (assign one package to the next available warehouse in round-robin order) or 'CLOSURE idx' (permanently disable that warehouse). Warehouses are skipped if closed or full. If a full round-robin scan finds every active warehouse is full, reset all active warehouses to their original capacities and continue assigning. Track how many packages each warehouse handles. After all operations, return the index of the warehouse with the most packages handled, breaking ties by largest index.

“The round-robin part was fine, I've done similar things.”

View Post

Capital One·Data Scientist·Technical Phone Screen

May 2026

Capital One data science interview with a pretty intense product analytics question that blended experimentation design with funnel thinking. One question, but it had a lot of moving parts.

How would you treat team matching as a funnel, define its key metrics, set weekly targets, and design a two-week experiment to improve match rate? Walk through your success metrics, minimum detectable effect, sample size assumptions, decision boundaries, and what you'd do if early results are inconclusive.

“This one had way more layers than I expected.”

View Post

Capital One·Software Engineer·Technical Phone Screen

May 2026

Capital One technical screen, one algorithmic problem the whole time. The problem was dressed up as a grid traversal thing but it's basically just DP with some expression parsing layered on top. Felt manageable once I stopped overthinking the operator handling.

Given an n x m grid where cells contain digits or +/- operators, find the maximum value of any path from the top-left to the bottom-right cell, moving only right or down. The value of a path is the arithmetic result of reading the digits and operators along it in order.

“I went straight to DFS and it took me an embarrassingly long time to realize memoization on just (row, col) might not be enough depending on how you handle operator state mid-path.”

View Post

Capital One·Data Scientist·Technical Phone Screen

May 2026

Capital One data scientist interview that goes deep on whatever ML algorithms you put on your resume. Not a vibe check, they actually want to know how the stuff works under the hood.

How does XGBoost handle missing values when building trees?
What regularization parameters does XGBoost expose, and how would you use them to reduce overfitting?
How does XGBoost compare to random forest, vanilla gradient boosting, and LightGBM?
What loss functions does XGBoost support, and how does using both the gradient and hessian speed up training compared to first-order methods?

“I knew the general answer but fumbled the wording.”

View Post

Capital One·Data Scientist·Technical Phone Screen

May 2026

Capital One data scientist interview that was basically one long technical problem about merging messy CSV files on a memory-constrained machine. Pretty deep in the weeds for a single question, covering everything from encoding detection to chunked streaming to ISO week math.

You're given four CSV files with inconsistent delimiters, mixed encodings, and possible duplicate header rows. On a laptop with 8 GB RAM and no cloud access, how would you merge them, handle files exceeding 10 million rows, and produce two specific output files: one with minimum purchase price per category (filtered by star rating) and one with weekly watch time per category?

“This was a beast.”

View Post

Capital One·Data Scientist·Onsite - Product Sense / Strategy

May 2026

Capital One data scientist interview with a single massive multi-part question that basically asked you to run a company for 45 minutes. The depth required was pretty wild for a DS role.

Pick a consumer app you like and break down its monetization model along with its top two competitors and what sets each apart.
As CEO of that app, define three north-star metrics and five guardrail metrics. Include precise formulas and specify which user events count toward each.
Propose one product improvement for that app with a falsifiable hypothesis, then design the full experiment including randomization unit, invariant checks, minimum detectable effect, statistical power, sample size, test duration, novelty effect mitigation, and seasonality controls. Walk through how you'd make the final launch decision.

“I picked Spotify because I actually use it and felt like I could talk about it without faking enthusiasm.”

View Post