System design round, implement a file system / tree crawler. You are provided a getPath method which can take a long time / fail when called. At the beginning you'd be asked to implement this in a single thread. Then, make it faster by implementing it in distributed manner. What happens if getPath fail / take a long time? How do you know when to stop crawling? What happens when you have many companies that you need to traverse? What database would you use? How would the end user fetch the result from the crawl? Retryable vs non-retryable errors?

Implement a JSON parser. You are given a list of tokens representing a JSON, like start-object, string, number, end-object. Then, implement nested objects and then implement array (i.e given start-array, end-array tokens). How to figure out when invalid token sequence is encountered? How to make your code cleaner.

Implement pivot table. You are given a table that tracks shirt sales. You want to perform aggregation on shirts sold where the columns and rows are colors and dates. Then you need to implement total rollup. Then how to aggregate across more than two columns.

For behavioral round. Conflict resolution. What project you are proud of and most impactful? Tell me about a time when you failed.