Capital One · Primly Community

Capital One coding interview / online assessment format and difficulty, 2026 update

mobile_mara · 5 replies

applied for a mid-level SWE role (mobile, but they interviewed me generically) and just finished the OA + phone screen coding. documenting this because i couldn't find recent info when i was prepping.

online assessment format: hackerrank-based. two coding problems, 90 minutes total. in my attempt it was: one medium-difficulty graph/BFS problem (finding shortest path with some constraints) one medium string manipulation problem with edge cases

not leetcode hard. nothing exotic. standard competitive programming patterns. if you can do 50-70 medium leets you're fine. the constraint was that you had to write production-quality code, not just get the right answer. they explicitly said they'd look at code quality and not just test pass rate.

phone screen technical (after OA): one problem, 45 minutes, with a recruiter-tech-hybrid interviewer. mine was a sliding window problem. they wanted clean code and asked me to walk through test cases out loud. they stopped me at one point and said "how would you handle null input" which almost tripped me up because i'd been moving fast.

one thing i noticed: they use a custom coding environment, not a shared Google doc or coderpad. you can't run the code locally. get used to talking through your logic more than running it.

difficulty vs other finance tech OAs: softer than Stripe or Jane Street. harder than most regional bank OAs. about on par with Fidelity and American Express from what i've seen.

if anyone has more recent data on whether they've changed the OA format in 2026, drop it below. i've heard they rotate prompts more aggressively now.

5 replies

jp_newgrad

the null input catch is so classic. they always do that. i've started literally writing "null/empty check" as a comment at the top of every function before i start just so i don't forget

bootcamp_bri

was there a time limit per problem or one shared 90-min timer across both?

mobile_mara

shared timer across both. technically you could spend 85 min on problem 1. don't. i split roughly 40/50 and it worked fine.

newgrad_neil

thanks for the hackerrank heads up, i kept seeing people say coderpad and was worried it was different. what was the code quality stuff they looked for? like comments, or just clean variable names?

mobile_mara

clean variable names, no magic numbers, one function per logical unit. i didn't write any comments and that was fine. readability over cleverness.