Let me start with the technical screen: a fairly classic multi-source BFS problem — find the nearest source for every cell in a 2D grid. Medium difficulty, leaning toward the basic side.
Next is the onsite, four rounds in total.
Round 1 was system design. The first 15 minutes were spent talking about a recent project of yours, then it moved into the design question. The task was to design an app that shows a rating for each food item (aggregated into a 5-point scale) along with the corresponding posts, where posts can be liked. It also needed to support settling driver payouts at the end of each month based on delivery activity. Overall pretty standard, mainly testing data modeling and system decomposition.
Round 2 was code craft. This round required calling multiple APIs to aggregate data and return a result; the APIs were all mocked. You had to handle API call failures yourself, and you could implement a retry mechanism, testing both single-failure and multiple-failure scenarios. Another annoying part was that you had to write your own test cases, and there was a very long expected result — you had to verify a deeply nested, large object character by character before you could move on to the next step. This round was not a great experience overall; the interviewer's style felt kind of pressuring.
Round 3 was the hiring manager round, fairly short at around 20 minutes, mainly talking about the biggest project you've worked on.
The last round was debugging. They gave a piece of load balancer code, and you needed to find and fix several bugs in it, then further optimize the routing strategy, e.g. implement round robin. Beyond that you could keep optimizing — I mentioned consistent hashing, but I didn't actually write it out.
Overall the questions were pretty standard, but the code craft round was a poor experience — the interviewer seemed to deliberately apply pressure, which affected my performance.
Discussion
Loading comments…