PayPal Data Scientist interview questions

53 Data Scientists have shared notes from their PayPal interviews, with the questions they were asked in each round. That is 58% of the 91 PayPal interview notes on Screna AI.

53 Data Scientist interview notes

PayPal·Data Scientist·Technical Phone Screen

Jul 2026
PayPal data scientist round focused entirely on a streaming anomaly detection problem. Pretty deep systems-thinking question for a DS role, felt more like a software engineering interview at points.
  • Design and implement a streaming login event processor that fires an alert whenever a user logs in from 4 or more distinct devices within any 10-minute sliding window. The function consumes a time-ordered stream of (user_id, timestamp, device_id, success) events. Walk through your data structures, eviction strategy, how you handle late-arriving events (up to 30 seconds), and the exact alerting semantics at window boundaries. Provide pseudocode.

“This was the whole interview basically.”

View Post

PayPal·Data Scientist·Technical Phone Screen

Jul 2026
Got a take-home style SQL case for a Data Scientist role on PayPal's fraud side. The problem was dense, like genuinely one of the harder SQL prompts I've seen in an interview context, and it tested whether you actually understand window functions and time-based filtering or just know how to write basic joins.
  • You're on a fraud detection team. Write a single SQL query that identifies candidate account takeover events in the last 7 days. A candidate event requires: the user's first successful login from a new device or new IP in the prior 30 days, followed within 2 hours by a transfer over $500 to a first-time recipient. Exclude users whose accounts are less than 7 days old at login time. If multiple transfers match the same login, keep only the earliest. Return specific output columns including a reason code, and at the end recommend indexes to support the query on a PostgreSQL-like warehouse.

“This one took me a while to decompose.”

View Post

PayPal·Data Scientist·Technical Phone Screen

Jul 2026
PayPal fraud data science interview that was basically a deep dive on p-values and experimentation. More rigorous than I expected for what was framed as a stats refresher.
  • Give a precise definition of a p-value. What does it actually mean, and what are common misconceptions about what it does NOT mean?
  • You run an A/B test on a new friction step like an extra OTP verification and get p=0.03. What can you actually conclude, and what else do you need before making a decision?
  • What are at least four common pitfalls when using p-values in practice? Think about things like multiple testing, peeking at results early, p-hacking, and so on.
  • How would you adjust your analysis when testing across many regions or segments, when outcomes are rare and delayed like chargebacks that arrive weeks later, or when randomization isn't clean due to spillover effects?

“I started with the textbook definition and then fumbled a bit trying to articulate the 'does not mean' part without sounding like I was reciting a Wikipedia warning label.”

View Post

PayPal·Data Scientist·Technical Phone Screen

Jul 2026
PayPal data scientist interview that leaned pretty hard into SQL fundamentals, specifically around aggregation and filtering. The question felt deceptively simple at first but there were a few layers they wanted you to work through.
  • Write a query to find merchants with a chargeback rate above 0.5% over the last 30 days using HAVING, and explain why you can't use WHERE for this.

“I knew the answer but fumbled explaining the WHERE vs HAVING distinction out loud.”

View Post

PayPal·Data Scientist·Technical Phone Screen

Jul 2026
PayPal data scientist experimentation round had a stats check that felt more like a quiz than a conversation. Quick binomial probability stuff, nothing that required much back-and-forth.
  • A fair coin is flipped 10 times. Calculate the total number of possible outcome sequences, the probability of exactly 3 heads, and the probability of at most 3 heads.

“Multi-part but not actually hard once you remember it's just binomial.”

View Post

PayPal·Data Scientist·Technical Phone Screen

Jul 2026
PayPal data scientist interview with a SQL-heavy technical screen. The question was specific and had a few layers to it, more than I expected for what I thought would be a standard analytics round.
  • Write a SQL query that uses CASE WHEN to produce conditional aggregates, such as counts of approved vs declined transactions per merchant and the sum of amounts flagged for review. Then explain why CASE WHEN is more portable across SQL dialects than approaches like SUM(column = 'x'), and walk through the readability and maintainability trade-offs.

“The query itself wasn't too bad, I got the structure right.”

View Post

PayPal·Data Scientist·Take-home Assignment

Jul 2026
PayPal DS take-home before onsite: you get a dataset, analyze an A/B test, and present slides. The scope was wider than I expected and the noncompliance angle tripped me up a bit.
  • What primary metric and guardrail metrics would you use to decide whether to ship the experiment?
  • Walk through the statistical tests and confidence intervals you would run, and what sanity checks you would perform before trusting any results.
  • How would you handle the fact that some assigned users never actually saw the new experience, and which estimand would you report?
  • How would you investigate heterogeneous treatment effects across platform, country, and user type without running into p-hacking issues?
  • What would your slide deck look like, and how would you communicate uncertainty and next steps to a PM or engineering audience?

“I went with orders_7d as primary and flagged revenue_7d and support_tickets_7d as guardrails.”

View Post

PayPal·Data Scientist·Hiring Manager Screen

Jun 2026
PayPal data scientist interview that leaned heavily on past experience in risk and fraud analytics. The main ask was basically a structured career walkthrough with a strategic twist at the end about why you'd want a role that's lighter on modeling.
  • Walk me through two or three past projects in risk or fraud analytics where you owned key decisions, moved measurable metrics, and had to navigate real trade-offs.
  • This role is less focused on modeling and more on analytics and strategy. Given your data science background, how does that shift align with what you're actually good at and what you want to be doing?

“This is the kind of question where you think you're ready and then halfway through the second project you realize you've been vague about what you actually decided versus what the team decided.”

View Post

PayPal·Data Scientist·Technical Phone Screen

Jun 2026
PayPal data scientist interview with a meaty unsupervised learning scenario that made me realize how much I'd been leaning on labeled data my whole career. No softball warmups, just straight into a hard problem.
  • You have millions of historical transactions with no fraud labels. Which unsupervised methods would you use to surface suspicious activity, and how would you evaluate the model's performance without ground truth?

“This one hit different because evaluation without labels is genuinely hard and I spent too long on the modeling side before they nudged me toward the second half.”

View Post

PayPal·Data Scientist·Onsite - Cross-functional / Panel

Jun 2026
PayPal fraud team interview for a Decision Scientist role on the Chicago ATO side. The whole thing was a dense stakeholder/BI case that covered dashboards, exec comms, and tooling debates all at once. More to unpack than I expected going in.
  • Walk through a 30/60/90 day plan for your first quarter on a fraud team, with specific deliverables at each milestone.
  • Define five core dashboard tiles for an account takeover monitoring dashboard, including the exact metric formulas and the right denominators.
  • Write a short executive update (around 150 words) to cross-functional stakeholders explaining a proposed fraud rule change, its expected impact, and the assumptions behind it.
  • Someone pushes back saying SQL is all you need for analysis. How do you decide when to use Python instead, and what governance practices do you put around code-based work to keep it trustworthy?

“I've done 30/60/90 questions before but never with this level of specificity demanded.”

View Post