Oracle · Primly Community

Oracle senior / L5 system design interview, what to expect

qa_quinn · 4 replies

went through the oracle senior swe loop earlier this year, austin office, OCI team. the system design round was not what i expected from an enterprise company.

first thing: they call it "L5" internally but their levels don't map cleanly to FAANG. oracle senior is roughly equivalent to L5 at Google or E5 at Meta in scope, but the interview depth on system design is actually pretty demanding.

what the round looked like: 60 minutes, one interviewer (a principal eng on the team). no collaborative whiteboard, just a shared doc. question was design a rate limiter for a multi-tenant API gateway. classic question, but they pushed unusually hard on: multi-tenancy specifically: how do you isolate rate limits per customer when one customer has 10 orgs under them consistency tradeoffs: they wanted me to pick between token bucket vs sliding window and defend it for a distributed setup failure modes: what happens when your redis cluster degrades. do you fail open or fail closed, and what are the SLA implications either way

compared to meta/amazon system design, i'd say this was more ops-focused. they kept asking "what breaks at scale" not "how would you build this greenfield." makes sense given OCI is an actual cloud business competing with AWS.

they did NOT ask me to code during the system design round. it was pure design + discussion. the behavioral portion was separate (30 min, different interviewer).

depth expected: be ready to go 3 levels deep on any component you mention. if you say "put a cache in front of it" they will ask you exactly how the cache invalidation works under concurrent writes. don't say things you can't defend.

time management: i ran slightly long on the data model portion and didn't get to the monitoring section, which i think hurt me. budget your time.

note for 2026 candidates: they seem to be hiring more for OCI-adjacent infra roles than application-layer stuff right now. if your background is cloud infra or distributed systems, lean into that in every round.

4 replies

sre_sol

the fail-open vs fail-closed question is such a trap. i've seen people lose offers because they confidently say 'fail open' without even blinking, and the interviewer just slowly nods while writing 'does not consider availability/security tradeoff.' what level were you interviewing for exactly, and did you get the offer?

backend_bekah

senior / IC4 equivalent. and yes, got the offer, didn't take it (comp wasn't competitive vs what i had). i said fail-open with rate limiting degraded to a fixed fallback threshold, which seemed to land okay. the key was i also flagged the security implications unprompted.

market_realist

this matches what i heard from a friend who did the loop for oracle cloud infra in seattle last fall. very ops-focused. they wanted to know you'd actually run something in prod, not just design it on a whiteboard.

sec_sasha

the multi-tenancy isolation question is interesting from a security angle too. rate limiting is one thing but did they ask about tenant data isolation at all or was it purely a performance/reliability lens?