Scale AI New Grad SWE Web Crawler Phone Screen

They were given a map from each URL to the URLs connected to it. The graph contained a cycle.

The first task was to write a crawler. The follow-up was to make it multithreaded and explain where locks were needed. Python was allowed but wasn’t required. This was the only technical problem in the interview.

The starting URL, crawl limit, output format, and ordering rules weren’t shared. They were rejected after the round.