Cloudflare · Primly Community

went through the full Cloudflare SWE loop last month, here's what I noticed

infra_ines · 4 replies

just wrapped a cloudflare loop for a senior SWE role on their network infra team. sharing notes while they're fresh.

rounds (fully virtual, spread over two days): recruiter screen, pretty standard, 30 min technical screen with a current eng: one medium-ish leetcode problem plus general CS conversation systems design: design a global rate limiting system (obviously very on-brand for them) coding round: two problems, one algo, one more systems-y behavioral: two separate rounds, different interviewers

the systems design round was the most interesting one by far. my interviewer pushed back every time I gave a vague answer. said things like "okay but what happens when the data center in São Paulo loses connectivity mid-request?" the anycast stuff came up, they wanted to know if I actually understood how they route traffic, not just that they have edge nodes.

behavioral had a strong ownership theme. "tell me about a time you were responsible for something that failed" came up in both behavioral rounds. they want specifics, not softened narratives.

turnaround after onsite was about 9 business days for me. a bit slow but recruiter gave a heads up that debrief scheduling takes time.

one thing I underestimated: they care a lot about the actual Cloudflare product suite. spent 30 minutes on Workers, their Zero Trust stuff, and what problems the company is actually solving. I had done homework but the depth of the product questions still caught me a little off guard.

4 replies

remote_swe_42

9 business days is on the faster end for them actually. I've heard of 2+ weeks before debrief even starts. did they give you feedback on the systems design or just a pass/fail?

infra_ines

no real feedback, just a brief "the team felt strong about your design thinking" type line from the recruiter. nothing granular. par for the course I guess.

sec_sasha

the rate limiting system design question is almost a rite of passage at cloudflare at this point. every eng I know who interviewed there got some variation of it. token bucket vs sliding window, what breaks at 100k rps, consistency tradeoffs across regions.

sre_sol

"tell me about a time something you owned failed" in TWO behavioral rounds is very on-brand for a company whose product IS reliability. they're not trying to trap you, they just genuinely want to know how you operate when things go sideways.