AWS Site Reliability Engineer Interview Questions

I would first identify which metrics actually help engineers make decisions and which ones are just clutter. Then I would consolidate overlapping charts, remove unhelpful graphs, and add context like deployment markers or thresholds. The reason I choose this approach is that a dashboard should make problems easier to see, not harder. I would also think about separating operational dashboards from business dashboards if both are needed. If the interviewer wants more detail, I would explain how I would review the dashboard with the people who use it most.

I would check pod restarts, resource limits, node health, network policies, and service discovery first. Then I would compare the failures to deployment timing, autoscaling events, and workload spikes. The reason I choose this approach is that intermittent failures in container platforms often come from resource pressure or configuration mismatches. I would also think about logging, probes, and whether the issue is tied to one node group or one service version. If the interviewer wants more detail, I would describe how I would isolate the problem with metrics and traces.

I would start by identifying domain boundaries and extracting the parts of the monolith that have clear ownership and scaling needs. Then I would move one service at a time, keeping the monolith and new services interoperable during the transition. The reason I choose this approach is that big-bang migrations are risky and usually create more downtime than necessary. I would also think about data ownership, communication patterns, and observability across services. If the interviewer wants more detail, I would mention that I would only split where the benefits justify the extra complexity.

I would describe a change that reduced toil, automated a recurring task, or removed a common source of incidents. Then I would explain the impact on uptime, incident volume, or engineer time saved. The reason I choose this story format is that reliability improvements are often about systems and process, not just more people. I would also mention how I measured success so the improvement was not just anecdotal. If the interviewer wants more detail, I would connect the outcome to the team or customer experience.

I would define the recovery objectives first, including RTO and RPO, because they drive the DR design. Then I would decide how data is replicated, how failover happens, and how the system is restored if a region or environment is lost. The reason I choose this approach is that disaster recovery should be designed around the business tolerance for outage and data loss. I would also think about testing the recovery plan regularly, not just writing it down. If the interviewer wants more detail, I would explain how I would validate the plan with exercises.