Former FAANG, recently went through McKinsey's system design round for a Staff-equivalent SWE role on their internal platform team. Sharing notes because I couldn't find specifics anywhere.
First: McKinsey doesn't call levels L5 or IC4. Internally they have their own career ladder but the external title was 'Senior Software Engineer.' Based on conversations during the loop it seemed to map roughly to FAANG L5/E5 scope expectations.
The format: 60 minutes, one interviewer who was a principal-level engineer at McKinsey. No whiteboard, we used a shared Google Doc. They explicitly said the doc was just for communication, not graded on neatness.
The problem I got: design a near-real-time analytics platform that lets consultants query engagement data across client projects. I had to think about multi-tenancy, data isolation between clients (obvious given McKinsey's confidentiality needs), latency targets, and how you'd handle schema-on-write vs. schema-on-read.
The client confidentiality angle was deliberate, I think. They wanted to see if I'd surface it as a constraint rather than them having to prompt me. I did surface it and it opened up a good 15-minute conversation about security architecture.
What they cared about: You drive the conversation. They don't spoon-feed requirements. Tradeoffs explained out loud. Say why, not just what. Domain-specific constraints (in their case: consultant workflows, multi-tenancy, audit trails) Rough capacity math. Didn't need to be exact, but I had to demonstrate I'd done it.
What surprised me: the interviewer was genuinely curious, not evaluating from a checklist. They pushed back on one of my choices and we just had an engineering debate. It felt less performative than some FAANG system design rounds I've been through.
Total interview feedback came back in about 5 days. They apparently debrief quickly once all rounds are done.