Stripe Coding Interview: What Gets Asked

The Stripe coding interview is practical, not LeetCode. You work in a real IDE in your own language, run your code as you go, and look things up. Signature rounds include bug squash (fix bugs in an unfamiliar codebase) and integration (build against a real API). The prompts read like a slice of the actual job, with stacked, escalating requirements, and you write your own tests. The bar is correct, readable, well-tested code, not optimal big-O.

A Stripe coding round puts you in a real IDE, in whatever language you like, writing code you run as you go. No whiteboard. The prompts read like a slice of the actual job. The score moves on code that comes out correct, readable and well tested. Sprinting to an optimal answer is not what moves it. Stripe's own recruiters tell candidates that practicing LeetCode "shouldn't really be necessary," and they mean it: the loop is built to test something else.

Forget the dressed-up graph puzzle. A Stripe prompt is a wall of text describing some payments mess, and the first test is pulling the actual problem out of it. You parse real-world input, apply a business rule, and then the interviewer stacks a second requirement on top, then a third. You write your own tests as you go. This lesson covers how a round runs, where the bar actually sits, and the four round types the rest of the section takes one at a time.

How a coding round runs

Below Staff, a Stripe onsite is virtual and runs about five hours: roughly five rounds of 45 to 60 minutes. Most of those rounds are hands-on coding of one flavor or another, plus a system design round and a behavioral round with the hiring manager. You code in your own editor with screen share, or in a shared pad like CoderPad, in whatever language you pick. The code runs. You are expected to run it, see it work, and fix it when it doesn't.

The prompts are long on purpose. A typical one hands you a string of rate pairs like "USD:CAD:1.3, AUD:USD:0.7" and asks for a direct currency conversion. Then it asks for a multi-hop conversion through intermediate currencies, then the cheapest route. Another hands you an Accept-Language header and a set of supported languages and asks you to return the supported tags in preference order. The work is parse the input, apply the money rule, then extend it as the interviewer adds parts. Reading the prompt closely and asking the right clarifying questions is part of the score, because the prompt is built to bury the real requirement.

The Stripe angle

Write the tests before anyone asks. On at least one Stripe prompt the instruction is spelled out in the problem itself: alongside the function, "you should use tests to demonstrate that it's correct." Running your code and checking it yourself counts for a lot here, so treat the tests as part of the job, not a chore you tack on at the end. Two more reflexes pay off on every prompt. Money lives in integer cents, never floats, because a floating-point rounding miss is exactly the correctness bug Stripe is hunting. And pick a language that makes parsing painless. One candidate switched to a scripting language mid-interview just to chew through the input faster.

What the bar really is

So drop the pattern catalog and drill a different muscle. Read a messy real-world prompt, turn it into clean code, get the money rules right. Then prove it works with your own tests while you talk through what you are doing. Nobody is timing you. They want code a reviewer would sign off on.

The rounds, by emphasis

Stripe builds the loop around round types, not algorithm families, so a frequency table tells you nothing. There is barely anything to count: the rounds are not an algorithm grind. The weights below rank the rounds by how much each one drives the decision, based on how candidates describe them.

Practical Coding is the core. That parse-then-apply-then-extend shape shows up there, and in the online assessment some candidates get first. Bug Squash and Integration each get their own round, and neither has a real FAANG analog, which is exactly why they catch prepared people off guard. Bug Squash is the round candidates name most often as the one that beat them. API and Systems Design weighs heaviest on senior and platform roles; it is correctness-first, not scale-first, and idempotency sits at the center.

The Stripe angle

Bug Squash and Integration are where the prep gap is widest. Spend real time on both, because no amount of LeetCode touches them. Bug Squash drops you into an unfamiliar codebase, and the bug is often a genuine historical one from an open-source project in your language. It rewards a clear loop: reproduce the failure first, form a hypothesis, make a small targeted fix, add a regression test. Integration is a live work-sample. You clone a private repo, read real API docs, and ship a small working program against a payments-style API, with full internet access but no AI assistants. Neither round cares how clever you are. They care whether you can stay calm inside code and docs you have never seen.

A newer format: the AI-assisted round

A newer round type has started surfacing in candidate reports. You code alongside an AI assistant, and you are scored on whether you catch and correct what it gets wrong. How fast the assistant produces something is not what gets scored. Some candidates describe a backend loop that folds in a round like this; others describe a frontend task, roughly 90 minutes, built against an undocumented API with an assistant in the loop. Treat it as reported and not yet universal. Still, be ready for a round that hands you a chatty co-pilot instead of a blank file.

How these lessons are organized

Each lesson below takes one round type. It opens with a short read on how Stripe frames that round and the signals it rewards. Then it works through the canonical task as a runnable example, with a solution you would be comfortable defending in review. Where a round leans on a data-structures or algorithm idea, the deep walkthrough lives in the free Foundations course. That keeps these lessons on the part that is specific to Stripe.