Meta Software Engineer Interview Questions

I would use two pointers, one for reading and one for writing. Since the array is already sorted, duplicates appear next to each other, so I can compare the current element with the last unique element and copy it forward only when it is different. The reason I choose this approach is that it gives O(n) time and O(1) extra space. I would also mention whether the function should return the new length or the modified array, because that changes the interface. Edge cases include empty arrays and arrays with all identical values.

I would first clarify whether mentions are used in chat, comments, or both, because the delivery rules can differ. At a high level, I would detect mentions at write time, store them as part of the event, and push them into a notification pipeline that fans out to the relevant users. I would keep mention parsing lightweight so it does not slow down the main posting flow. The reason I separate mention detection from notification delivery is that the downstream fan-out and ranking logic can grow independently. I would also think about spam prevention, duplicate notifications, and offline delivery.

I would first think about sorting meetings by start time and tracking the rooms that are currently in use. A clean approach is to use a min-heap of end times so I can always reuse the room that becomes free first. Whenever the earliest ending meeting finishes before the next meeting starts, I reuse that room; otherwise, I allocate a new one. The reason I choose a heap is that it gives a good balance between clarity and efficiency. The time complexity is O(n log n) because of sorting and heap operations, and I would also mention edge cases like meetings that start exactly when another ends.

Before designing it, I would clarify what actions must work offline, how long data can stay stale, and how conflicts should be resolved once the device reconnects. At a high level, I would keep a local database on the device, queue user actions locally, and sync them with the server when the network returns. The reason I choose this architecture is that mobile users expect the app to remain usable even with poor connectivity. I would also think about conflict resolution, retries, and background sync because those are the tricky parts of real offline support. The trade-off is between consistency and usability, and I would usually prefer a design that keeps the app responsive and resolves conflicts predictably.

I would start by thinking about whether I can do it iteratively or recursively. The iterative version is usually easier to explain and more space-efficient: I keep track of previous, current, and next pointers, then flip the current node's pointer as I walk through the list. The reason I choose this approach is that it gives O(n) time and O(1) space. If the interviewer asks for recursion, I can also explain that version, but I would mention it uses extra stack space. I would also point out edge cases like an empty list and a single-node list.