did oracle's onsite last month for a senior SRE role, cloud reliability team. 'onsite' is a misnomer now, it was all video, four back-to-back 45-minute blocks on a single day with two 10-minute breaks.
block 1: technical deep dive. this was essentially a system design + architecture conversation tailored to reliability. they gave me a scenario: an oracle cloud service is seeing elevated error rates, latency spikes, and one region is seeing higher rates than others. walk me through how you'd investigate and respond. this is their version of an incident response simulation. they want to see: instrumentation instincts, where you look first, how you communicate during an incident, how you do a root cause write-up after.
block 2: coding / technical problem-solving. even for an SRE role, there was a coding segment. medium difficulty, wrote a function to parse and aggregate log entries by error code across time buckets. python or go, i used python. no leetcode, just practical scripting that an SRE would actually write.
block 3: behavioral / leadership. classic behavioral questions: describe a time you improved system reliability, a time you had to push back on a deadline for safety reasons, a time you worked with a dev team that kept breaking your services. they scored this pretty rigorously, i could tell the interviewer was taking notes on specifics.
block 4: hiring manager conversation. not a 'let's see if you'll be a culture fit' chat. the hiring manager had clearly read the notes from the other three rounds and asked pretty pointed follow-ups. they also explained the team's roadmap and current challenges. felt more like a real conversation than an evaluation.
biggest surprise: the level of specificity expected. oracle doesn't want vague incident stories. they want: what was the error rate, what was the latency p99, how long did remediation take, what was the postmortem action item. numbers throughout.
note: the overall day took about 5 hours. i was drained. pace yourself.