Explain Adjacency List. What are its pros and cons?

0-2 years experience

1. 1We need to store a small social network where each user follows a few others. How would you represent this using an adjacency list in code?
2. 2If you have a graph with five vertices and you add an edge between vertex 2 and vertex 4, what does the adjacency list look like after the insertion?
3. 3What happens when you iterate over the neighbors of a vertex that currently has no edges?

2-5 years experience

1. 1Your pathfinding feature uses an adjacency list, but after a data import some nodes have duplicate neighbor entries. How would you detect and clean those duplicates?
2. 2Checking whether an edge exists between two vertices is slower than expected. Why might an adjacency list cause this, and what alternative representation could you consider?
3. 3We need to support both directed and undirected graphs with the same codebase. How would you adapt your adjacency list implementation to handle both cases without duplicating logic?

5-8 years experience

1. 1You are designing a microservice that stores a massive road network (millions of vertices, billions of edges). Explain how you would store the adjacency list on disk and the trade‑offs for read vs. write performance.
2. 2A memory leak appears when repeatedly adding and removing edges in a long‑running analytics job. Walk through how you would investigate and fix the leak in an adjacency‑list implementation.
3. 3When scaling the graph to a distributed environment, what challenges arise with an adjacency list, and how would you partition the data to minimize cross‑machine edge lookups?

8+ years experience

1. 1Our platform is migrating from an in‑memory adjacency list to a persisted graph store. What architectural considerations would you evaluate to ensure backward compatibility and minimal downtime?
2. 2Different teams use adjacency lists for real‑time recommendation and batch analytics. How would you design a shared graph service that serves both low‑latency edge queries and high‑throughput traversals?
3. 3If we need to evolve the graph schema (adding edge types, timestamps) while keeping existing adjacency list code functional, what strategy would you propose for versioning and migration?