Uber Software Engineer Interview Questions

I would decide what identity to limit on, define the allowed burst behavior, and choose a storage strategy that works across multiple instances. Then I would make the system accurate enough to protect the service without blocking normal traffic. A good answer also explains how to observe and tune the limiter over time.

I would model the map as a graph and choose a shortest-path algorithm based on the cost structure. If edge weights matter, I would use something like Dijkstra’s algorithm. The key is to match the algorithm to the shape of the problem.

I would separate ingestion, storage, and read paths so the system can handle frequent updates cleanly. Then I would think about freshness, geospatial lookup, and how to keep updates efficient for both drivers and riders. The main challenge is balancing speed with accuracy at scale.

I would use a standard geometry approach such as ray casting or winding number, depending on the exact requirements. Then I would make sure edge cases like points on boundaries are handled clearly. A strong answer shows that you can turn a geometric idea into correct code.

I would start with the user flow, the main entities, and the biggest scale bottlenecks. Then I would design for availability, latency, and fault tolerance around those core flows. The goal is to show that you understand both product behavior and engineering constraints.