"Tell me about a time you noticed an operational inefficiency in your infrastructure or developer workflow and fixed it without being asked. For example, this could be in CI/CD, oncall, capacity management, or internal tooling around systems like Sandcastle or Scuba. Walk me through how you identified it, why you chose to act, what you did, and what changed as a result."
This question tests ownership beyond assigned tickets. For a DevOps Engineer at Meta, interviewers want to see whether you can spot friction in large-scale systems, prioritize the right problem, and improve reliability or velocity without waiting for formal direction. They are also looking for influence without authority: many inefficiencies span multiple teams, so fixing them often requires alignment, not just technical execution.
A strong answer uses one specific example with clear stakes, shows how you validated that the inefficiency was real, explains the trade-offs you made, and quantifies the impact on reliability, developer productivity, or incident load. The best answers also show how you brought others along, avoided over-engineering, and learned something that changed how you operate afterward.