LinkedIn Software Engineer Interview Questions

I would split message ingestion, delivery, storage, and presence so the system can scale each part independently. Then I would think about ordering, retries, offline delivery, and read receipts. The key is to make the system reliable even when users are online in different ways.

I would use depth-first search with a visitation state that tracks nodes currently on the recursion path. If I encounter a node that is already on that path, there is a cycle. That gives a clean and efficient way to reason about graph structure.

I would think about partitioning, replication, consistency, and recovery first. Then I would decide how data is placed, how failures are handled, and what guarantees the store should give to clients. A strong answer shows you understand the trade-offs between simplicity, speed, and durability.

I would ask clarifying questions first so I understand scale, latency, consistency, and user goals. Then I would make reasonable assumptions and call them out clearly before designing. Interviewers usually care more about how you think than whether you guessed every requirement correctly.

I would use a hash map for direct access and a doubly linked list for recency order. That lets me update and evict items in constant time. The design is popular because it is both fast and straightforward.