Social graph traversal, feed ranking and search relevance - the domain shapes almost every design question.
LinkedIn design rounds are graph and feed rounds. Expect questions about traversing a social graph at scale - degrees of connection, people you may know - and about feed generation, where the fanout-on-write versus fanout-on-read trade-off is the central decision and the follow-up is always what you do about the users with millions of connections.
Search relevance appears frequently too: retrieval, ranking features, personalisation, and how you evaluate whether a ranking change actually helped. The coding rounds are standard big-tech fare at medium to hard difficulty, and the behavioral round leans on collaboration and mentorship rather than a proprietary values framework.
45 min
Medium to hard algorithmic problem.
45 min each
Two rounds, graphs and heaps well represented.
60 min
Feed generation, graph traversal at scale, or search relevance.
45 min
Collaboration, mentorship and ownership.
Design the feed. The decision that matters is fanout on write versus fanout on read, and the follow-up is always the hub problem - what you do for accounts with millions of connections, where pure write fanout is untenable.
Enough to reason about traversal cost at scale. Degrees of separation, why a naive BFS over a hot node is expensive, how you store and shard graph edges, and what you precompute versus compute on request.
Standard big tech: medium to hard, with graphs and heaps over-represented relative to other companies because of the domain.
Practise the rounds LinkedIn actually runs, and get scored on where you stand.