Just finished the full loop for a staff-ish / senior SWE role at Coca-Cola's Global Digital Technology group. The system design round surprised me, so I'm writing this up while it's fresh.
The prompt: Design a real-time inventory and order management system that handles Coca-Cola's digital sales channels (think their e-commerce, retailer APIs, DSD delivery scheduling). They were vague on exact scale but when I asked, they gave me numbers: ~50,000 SKUs, ~2 million orders per day peak, global but US-first.
What they cared about: How you handle eventual consistency between warehouse systems and the digital storefront. they pushed on this. Your approach to failure modes. what happens when the retailer API is down? do you retry, queue, degrade gracefully? API design between microservices. they asked whether I'd use REST or event-driven for specific flows and wanted justification, not a default answer. Observability. one interviewer asked how I'd know if something broke silently. that's an SRE-adjacent question and I appreciated it.
What they didn't really care about: Database-level sharding math. I mentioned partitioning strategies but they didn't drill into the byte arithmetic. Trendy tech names. I mentioned Kafka for the event stream and they nodded, but they would have accepted RabbitMQ or Pub/Sub. They wanted the reasoning, not the brand.
My take: this is a genuinely senior-level system design question. It's not "design Twitter" territory but it's also not a junior warm-up. If you're coming in at the senior / L5 equivalent level, expect them to probe on distributed systems fundamentals, not just high-level boxes and arrows.
One tactical thing: they shared a virtual whiteboard (Miro-like tool). Draw your components early and narrate as you go. I noticed the interviewers relaxed when I had something visual to point at.
Happy to talk through the specific design I used if anyone's prepping for this round.