Did the CrowdStrike senior SWE loop earlier this year. The system design round deserves its own post because it's genuinely different from what you prep for at consumer tech companies.
First: they don't call levels "L5" internally, but the senior IC band maps roughly to what the industry calls that. If your recruiter uses "Senior Software Engineer II" that's the one.
The prompt itself. Mine was something along the lines of: design a real-time threat detection pipeline that processes endpoint telemetry at scale, catches anomalies, and routes high-confidence detections to analyst queues. Very on-brand. Your mileage will vary but expect a security-domain problem, not a ride-share or social feed.
What they actually evaluated: How you scope ambiguity. I asked about SLA requirements for detection latency and the interviewer lit up. That was the right move. Data flow clarity. Draw the full path: agent on endpoint -> transport -> ingestion -> stream processing -> storage -> alerting. Know each handoff. Trade-off reasoning. I went Kafka for the message bus. They asked why not a simpler queue. Have a real answer. Scale numbers. They asked me to estimate events-per-second from a fleet of 100k endpoints. Back-of-envelope math, don't skip it.
What they didn't really care about. UI layer, exact database schema for logging, getting the perfect architecture on the first try. This is systems-heavy, not API-design-heavy.
Prepared by reviewing distributed stream processing concepts and reading about endpoint detection pipelines. That last part helped a lot. If you're interviewing for the sensor or detection side of the house, understanding how EDR data flows will make you sound way more credible than generic prep.
One more thing: the interviewer was a principal engineer. She pushed back on two of my choices. That's not a red flag, that's how it goes at this level. Pushback is the signal, not a problem.