Meta Backend Engineer Interview Questions

I would start with the failing payload, logs, and stack trace so I can reproduce the issue in a smaller environment. Then I would figure out whether the failure comes from bad data, a code bug, a timeout, or a race condition. The reason I start there is that worker failures often come from a single edge case rather than the whole pipeline. I would also think about whether the job should retry, skip, or go to a dead-letter queue after repeated failures. Once I find the root cause, I would add validation or stronger error handling so the same input does not break the system again.

I would separate file bytes from metadata. The file content would live in object storage, while metadata like owner, name, size, and permissions would go into a database. For large uploads, I would support chunked or resumable transfers so users do not have to restart a failed upload from scratch. The reason I choose this split is that it makes the system scalable and keeps lookup and authorization simpler. I would also mention signed URLs, antivirus scanning, and CDN support for popular files.

I would assume duplicates are possible because retries and consumer restarts are normal in distributed systems. To make processing safe, I would make the consumer idempotent and store a record of processed message IDs or final results. The reason I choose this approach is that it is usually easier to make the work safe than to eliminate every duplicate at the transport layer. I would also think about how long to keep deduplication records and how to handle jobs that fail halfway through. If needed, I would explain the difference between at-least-once and exactly-once processing.

I would first define what counts as activity, because it could mean requests, messages, time spent, or something else. My design would stream events into an aggregator, keep counts for the current window, and maintain the top 10 with a small ranking structure. The reason I choose a streaming approach is that it avoids recomputing everything from scratch every minute. I would also talk about sharding, merging partial aggregates, and handling late events. If freshness matters, I would keep the aggregation pipeline low latency.

I would put a rate limiter or gateway protection in front of the API. My first choice would be a token bucket because it allows bursts while still enforcing a long-term cap. The reason I choose this approach is that it protects the backend without making the system too strict for normal traffic spikes. I would also discuss whether limiter state should be shared across instances and what should happen if the limiter store is unavailable. Related protections like load shedding and circuit breakers would also be worth mentioning.