← Back to Directory

Capital One interview notes, page 1 of 39

Large Enterprises

Most of these Capital One write-ups come from Data Scientists (196), then Software Engineers (93) and Product Managers (54). The most reported rounds are the technical phone screen (213), the product sense round (58), the behavioral / leadership round (41) and the system design round (22).

382 interview notes · updated Oct 2026

Capital One·Data Scientist·Onsite - Cross-functional / Panel

Feb 2025
Cleared a company-wide panel at Capital One in late February for a Data Scientist role, then hit a wall during team matching. Three separate team meetings over the course of a week and by Friday the recruiter had nothing, both teams were still talking to other candidates.
  • How would you design a 14-day follow-up plan to maintain momentum with multiple recruiters after team-match meetings stall, without coming across as pushy?
  • At what point in a stalled hiring process should you escalate to a more senior contact, and what criteria would you use to make that call?
  • How do you decide when to broaden your outreach to other teams or companies versus continuing to wait on a process that has gone quiet?

“This is basically a stakeholder management problem dressed up as a job search question.”

View Post

Capital One·Data Scientist·Technical Phone Screen

May 2025
Capital One Data Scientist interview with a dense multi-part SQL question. One problem, but it covered basically everything: CTEs, NULLs, aggregates, UNION ALL, conditional logic. More of an exam than a conversation.
  • Given a student/exam/scores schema, write a single SQL statement (CTEs allowed) that: computes subject-level metrics for May 2025 including a floored average, a high-scorer count, and a low-score percentage (handling NULLs and subjects with zero attempts); labels each student-subject pair with a pass or remedial flag based on their best May score; and combines both result sets with UNION ALL, adding a level column and NULLs for inapplicable columns.

“This one wrecked me a little.”

View Post

Capital One·Data Scientist·Technical Phone Screen

Aug 2025
Capital One data scientist interview that was basically a single massive case study about a two-sided marketplace with a conversion drop. The question had five sub-parts and covered everything from metric decomposition to experiment design to power calculations. Dense stuff.
  • A booking conversion rate dropped from 3.2% to 2.4% after a ranking change shipped. Build a metric tree from conversion down to search, view, contact, and booking rates. Which slice-and-dice checks and counter-metrics would you run first to separate cause from correlation across supply elasticity, position bias, latency, partner acceptance, and cancellations?
  • Propose a log-based analysis to attribute the conversion drop to the ranking change versus supply or traffic mix confounders. Think about approaches like diff-in-diff across unaffected suburbs, CUPED, or synthetic controls.
  • Design an A/B test for a candidate ranking fix. Define the primary metric, guardrail metrics like cancellations, partner rejection rate, and search latency p95, and describe how you would pre-register the experiment.
  • With a baseline CVR of 2.4%, a target relative lift of 6%, a 50/50 split, 14-day runtime, 200k daily sessions, alpha 0.05, and power 0.8, are you sufficiently powered? If not, what levers would you adjust, and how do you handle marketplace interference and SUTVA violations?
  • If your experiment results are heterogeneous across city, device type, and traffic channel, what would you actually ship?

“This is where I spent most of my mental energy and still felt like I left stuff on the table.”

View Post

Capital One·Data Scientist·Technical Phone Screen

Sep 2025
Capital One DS interview with a meaty A/B testing case study centered on an ad scheduling policy experiment. The whole thing was basically one long scenario with sub-questions layered on top of each other, which I wasn't expecting for a phone screen format.
  • What should be the randomization unit for this experiment: user-level, device-level, or impression-level? How would you prevent cross-contamination across platforms and time slots, and what hashing or bucketing key would you use?
  • How would you stratify the experiment and apply variance reduction? Specify the strata, whether you'd use CUPED, the exact pre-period dates, the covariate definition, and the adjusted estimator formula.
  • Formally define the primary KPI and guardrail metrics, including numerators and denominators. Will you use intent-to-treat? How do you handle zeros and outliers, for example through winsorization?
  • Should the hypothesis test for the primary KPI be one-tailed or two-tailed, and why? Which test would you use and what is the test statistic?
  • Outline the sample size calculation for detecting a 5% relative lift in mean watch time per impression, given a baseline mean of 42 seconds, standard deviation of 55 seconds, alpha of 0.05 two-sided, and 80% power. Also describe any sequential monitoring approach if you plan interim looks.
  • Define the 95% confidence interval you would report, how you would pool results across strata, and how you would adjust for multiple guardrail tests. Also describe at least two randomization balance checks and two seasonality or novelty effect diagnostics.
  • How would you adjust the analysis if offline conversion data is incomplete for the last 48 hours of the experiment window? How does that affect the readout?

“I went with user-level pretty quickly and felt okay about that, but fumbled a bit explaining the bucketing key.”

View Post

Capital One·Software Engineer·Technical Phone Screen

Apr 2026
Capital One technical screen focused entirely on matrix manipulation. One problem, five operations, and a lot more edge cases than I expected going in.
  • Given an m x n integer matrix, implement five operations: swap two rows, swap two columns, reverse a given row, reverse a given column, and rotate the entire matrix 90 degrees clockwise. Keep in mind the matrix may not be square, so rotation must produce an n x m result.

“The first four operations were fine, pretty mechanical.”

View Post

Capital One·Product Manager·Onsite - Product Sense / Strategy

Apr 2026
Capital One PM interview, one question about growing a specific product feature. Pretty focused session, felt more like a product strategy exercise than a broad loop.
  • How would you grow the virtual card number feature?

“I went straight into user segments and almost forgot to anchor on why people use virtual cards in the first place.”

View Post

Capital One·Software Engineer·Technical Phone Screen

Apr 2026
Capital One research engineer interview with a graph/optimization problem that took me a while to wrap my head around. The drone delivery problem sounds applied and cool until you realize it's basically a shortest path variant with constraints.
  • Design an algorithm to deliver a package from a start to a target location using drone flights and walking. Drones can only recharge at specific stations and have a limited range. Minimize the total walking distance the delivery person must cover.

“My first instinct was Dijkstra but I kept second-guessing whether to model walking distance or total distance as the edge weight.”

View Post

Capital One·Product Manager·Onsite - Product Sense / Strategy

Apr 2026
Capital One PM interview with a product strategy question that was deceptively open-ended. No fluff, just one meaty case and a lot of silence while I tried to figure out where to start.
  • How would you design a revenue curve for a new Capital One product?

“I fumbled the opening badly.”

View Post

Capital One·Product Manager·Onsite - Behavioral / Leadership

Apr 2026
Capital One PM interview, just the one question about a feature launch. Pretty standard but it's the kind of thing that sounds easy until you're actually in it trying to remember specifics.
  • Walk me through a feature you've launched from start to finish.

“I had an answer prepped but I fumbled the structure a bit, jumped straight into the solution without setting up the problem context first.”

View Post

Capital One·Machine Learning Engineer·Technical Phone Screen

Apr 2026
Capital One MLE interview that was basically a deep dive into ML fundamentals across the board. Six topics, all technical, no fluff. Felt like a written exam more than a conversation.
  • Explain the bias-variance trade-off: what each term means, how they contribute to total prediction error, and how model complexity and regularization shift the balance.
  • What's the difference between parameters and hyperparameters? Give concrete examples of each and explain how they're set.
  • Compare batch inference and real-time inference. How do they differ in latency, throughput, infrastructure, and when would you choose one over the other?
  • How would you handle training a model on very large datasets that don't fit in memory or on a single machine?
  • What is feature engineering, why does it matter, and what are the common techniques? How do you think about it given the move toward learned representations?
  • Walk through the full ML lifecycle from problem framing to retraining, and describe a real end-to-end project you personally delivered.

“This is one I've explained a hundred times but still fumbled the decomposition formula under pressure.”

View Post