← Back to Directory

Coinbase interview notes, page 20 of 21

Large Enterprises

Most of these Coinbase write-ups come from Software Engineers (122), then Data Scientists (41) and Product Managers (25). The most reported rounds are the technical phone screen (112), the system design round (31), the product sense round (25) and the online assessment (11).

210 interview notes · updated Jul 2026

Coinbase·Product Manager·Onsite - Product Sense / Strategy

Jul 2026
PM interview at Coinbase with a product sense question that felt oddly out of place for a crypto company. One question, pretty open-ended, left me second-guessing my framing the whole time.
  • How would you optimize Booking.com to improve its conversion rates?

“Weird question to get at a crypto company but okay.”

View Post

Coinbase·Software Engineer·Technical Phone Screen

Jul 2026
Coinbase frontend interview where they had me build a full autocomplete component from scratch. Pretty involved for a single session, covering debouncing, async fetching, keyboard nav, and styling the active state all at once.
  • Build an autocomplete input component that debounces user input, fetches suggestions asynchronously, renders the list, supports up/down arrow key navigation with the active item highlighted, and submits the selected value on Enter.

“This was a lot to juggle in one question.”

View Post

Coinbase·Data Scientist·Technical Phone Screen

Jul 2026
SQL-heavy technical screen for a DS role at Coinbase. Three questions, all back-to-back, all on the same invented schema. No behavioral stuff at all which surprised me, just pure query writing under time pressure.
  • Given a users table and a transactions table, write SQL to compute adoption rate and transaction rate per region for August 2025. Adoption rate is the share of users in that region who adopted in August, transaction rate is the share who made at least one transaction in August.
  • For each user, calculate the number of days between their adoption date and their first transaction. Only include users where adopted_at is not null and the first transaction happened after adoption. Then write a second query returning the p10, p50, and p90 of that value.
  • Define a cross-region sale as any transaction where the transaction region differs from the user's earliest-ever transaction region. Write a query that returns all August 2025 transactions with a flag indicating whether each is a cross-region sale, plus the user's first transaction region.

“Took me longer than it should have to realize I needed to keep all users in the denominator, not just the ones with activity.”

View Post

Coinbase·Data Scientist·Technical Phone Screen

Jul 2026
Interviewed for a Data Scientist role on the Identity & Trust team at Coinbase. The whole thing was essentially one long experiment design question with a lot of moving parts, and I left feeling like I'd covered maybe 70% of what they were actually looking for.
  • How would you design an A/B test to evaluate the launch of an identity verification feature with trust signals like verified badges and suspicious profile warnings? Walk through randomization, metrics, rare event measurement, network effects, bias sources, and how you'd interpret mixed results.

“This was basically a gauntlet.”

View Post

Coinbase·Data Scientist·Technical Phone Screen

Jul 2026
Coinbase data scientist interview with a scenario-based question about diagnosing a metric anomaly in the wallet product. One question, but it had a lot of layers to it.
  • If you noticed an unusual change in engagement or a key metric for Coinbase Wallet, how would you structure your investigation? What hypotheses would you form, and what data would you use to confirm or rule each one out?

“This is one of those questions where the answer lives in how you organize your thinking, not in any single clever insight.”

View Post

Coinbase·Data Scientist·Technical Phone Screen

Jul 2026
Coinbase data science interview with a product analytics case built around a Super Bowl QR code campaign. The whole thing was funnel estimation, which sounds straightforward until you're actually in it trying to justify every assumption out loud.
  • Given a TV network airs a QR code ad during the Super Bowl, how would you estimate the percentage of viewers who actually scan the code?
  • What data sources would you draw on to support this kind of scan rate estimate?
  • If you only have prior TV campaign data and no ad-partner conversion data, how do you approximate the scan rate?
  • How would you estimate the conversion rate from someone scanning the QR code to completing an account sign-up?

“I went straight to viewership numbers and tried to layer in attention decay, second-screen behavior, that kind of thing.”

View Post

Coinbase·Data Scientist·Technical Phone Screen

Jul 2026
Coinbase data scientist interview that dropped a full end-to-end modeling exercise on me with no warm-up. You're basically doing a live analysis from raw data to model evaluation to recommendations, all on the spot. Pretty intense if you're not used to thinking out loud under pressure.
  • Given a raw behavioral dataset, walk through your full analysis pipeline: clean the data, explore it, build a model to predict conversion, evaluate it, and propose improvements.

“This was basically the whole interview compressed into one question.”

View Post

Coinbase·Data Scientist·Technical Phone Screen

Jul 2026
This was a deep-dive technical screen for a DS role at Coinbase, basically a full experiment design case with a debugging twist. The questions were meaty and required actual math, not just framework recitation. Left feeling like I'd run a marathon.
  • Define a precise primary metric and two to three guardrail metrics for a personalized product email experiment. Assume intent-to-treat with user-level randomization, and explain how you'd handle eligibility rules and users who receive multiple emails.
  • Given a baseline 7-day purchase conversion of 3.5%, calculate the sample size needed to detect a 10% relative lift at two-sided alpha 0.05 and 80% power with a 1:1 split. With 500,000 eligible users per day and 85% deliverability, how many calendar days does the test need to run including the full attribution window?
  • Now power the same experiment for 7-day revenue per randomized user instead of conversion rate. Mean is $0.90, SD is $12.00, and you want to detect a $0.10 absolute lift. What per-arm sample size and run length does that imply?
  • Compute the two-proportion z-test p-value and 95% confidence interval for each of the two experiment runs separately, then compute a fixed-effects meta-analytic pooled estimate across both. Based on that, should the team launch?
  • List at least six plausible reasons why the first and second experiment runs produced such different results. For each, describe the specific data checks or queries you'd run to investigate.
  • Propose a full re-analysis plan covering pre-registration, covariate adjustment, heterogeneity of treatment effects, sequential monitoring corrections, and a holdout strategy for ramp. What decisions would you make if the pooled lift lands between zero and five percent?

“I fumbled the multi-email part initially.”

View Post

Coinbase·Software Engineer·Technical Phone Screen

Jul 2026
Coinbase software engineer screen, one coding question the whole time. Pretty straightforward concept but the edge cases are where they actually want to see you think.
  • Design and implement a pagination helper class that takes an array and a page size, and returns the correct slice of items for a given zero-based page index. Handle out-of-range page numbers gracefully.

“Seemed easy at first and I jumped straight to the slicing logic, which was a mistake.”

View Post

Coinbase·Software Engineer·Technical Phone Screen

Jul 2026
Coinbase software engineer interview with a coding question around building a Bank account class. Pretty straightforward OOP problem but the edge cases around overdraft prevention and cashback logic are where things get interesting.
  • Design and implement a BankAccount class with Deposit, Withdraw, and CashBack methods. The class should prevent overdrafts and return the updated balance after each operation.

“Felt pretty confident going in since I'd done similar OOP stuff before.”

View Post