These mistakes fail candidates:
Jumping to solutions. You start drawing databases before understanding what you're building. Slow down.
No trade-offs. Every decision has downsides. If you can't name them, you don't understand the decision.
Silence. Your interviewer can't score what they can't hear. Think out loud.
Over-engineering. Adding Kafka, Redis, and microservices to a problem that needs database.
Under-engineering. Ignoring scale. "Just use a single server" doesn't work for million users.