Capital One Data Scientist Interview Questions

Explain and implement LoRA by learning a low-rank update to a frozen weight matrix.

model architectureDeep Learningmodel fine-tuningCapital One

Determine whether reversing a $1M stock position changes VaR, and distinguish VaR magnitude from the direction of the P&L.

financial calculationsRisk ManagementarithmeticCapital One

Explain how to build and interpret a ROC curve and confusion matrix, and how they map to business outcomes.

ClassificationConfusion MatrixAUC-ROCCapital One

Create a reusable preprocessing pipeline that encodes categories, imputes missing values, and standardizes training and testing data consistently.

Data Qualitydata processingETLCapital One

Approach for improving onboarding to a high-risk financial product while balancing conversion, compliance, and user protection.

onboardingUser NeedsConversion RateCapital One

Design a UI experiment that distinguishes short-term novelty lift from durable engagement impact under fixed traffic and guardrail limits.

ExperimentationStatistical SignificanceNovelty EffectCapital One