I've already forgotten what the phone screen was, but it wasn't hard.
The onsite loop was really different from other companies — felt different from what I'd seen and from other people's experiences too.
Round 1 was a bug-hunting round, roughly three problems. Make sure you set up your IDE and install the plugins ahead of time — the details are in the email. (I wasn't careful and didn't read the email properly, so the first half hour of this round was spent downloading the IDE and doing all the setup.) The problems are hard to explain, and I forget what they were, but they weren't hard — you just had to look at the test cases and start from wherever the error was thrown.
Round 2 was AI-assisted coding, simulating using Codex to solve a coding problem. I think the problem gave you two parameters: the first was a set of rules represented as strings, like "a credit card name can't equal X" or "a merchant name can't equal Y"; the second parameter was, I think, an array of strings representing transaction records. It was basically about how you use the Codex agent to write something that passes the test cases. You had to write the test cases yourself too.
Round 3 was coding without AI. The problem gave you JSON data — a series of x/y coordinates — and you had to use an API to connect these points and draw them as a map. It's an open-source API you can actually call, and they give you a sample request body too.
Round 4 was the behavioral round — a manager talking with you about past projects, leadership, nothing special. I was interviewing for a senior role, so maybe they wanted a stronger leadership signal.
Weird thing is, I was interviewing for Sr but there was no system design round?! And no traditional LeetCode-style problems either.
About 2-3 days later they told me I didn't get it. I definitely failed the bug-hunting round, and the other three rounds were pretty mediocre too.
Discussion
Loading comments…