First round phone screen:
I picked AI coding, and it was the logger question that's been posted on the forum. Roughly: the first question was to implement the following functionality, where each feature could be triggered and then print out the logger:
- Removes all occurrences of a configured string before printing.
- Truncates the message to a max number of characters before printing.
- Capitalizes the entire message before printing.
- Stores messages in an internal list without printing.
The second question: within the "store" feature from part four of the first question, implement a search function. They didn't ask me to actually write it, just talk through it.
Three virtual onsite rounds:
AI coding: the camel card problem from the forum.
SD round: a Google News aggregator. They asked about things like the polling strategy for pulling data from a third-party API, and personalization only needed to cover two things — letting users follow publishers and pick their favorite categories.
Overall, I never really figured out what the AI coding round was actually testing. My usual approach is to talk through my ideas with the interviewer first, then feed them to the AI to generate an approach and code, then check whether what it wrote is correct, then paste it over to run the tests. I didn't really get why the interviewer said it's not recommended to paste the entire question straight into ChatGPT — but rephrasing it takes time and I might drop some of the question's constraints in the process. Both of my AI coding rounds had this problem, which ended up costing me some debugging time when I got to the tests.
Discussion
Loading comments…