AI PM comp at the top has gone absurd. At companies like OpenAI, Google, and Meta, yearly stock grants are running around $1.5M per employee. But the interviews have changed underneath it. The "design a fridge for blind people" era is over. The top AI PM interviews now test system design. Can you reason about models, data, memory, latency, and failure modes the way an ML engineer would? I ran a full mock interview with Aman on this exact question: "Build the system design for a churn…