This company still has an online OA, which I found a little surprising. It included some multiple-choice questions, and the last problem was a divide-and-conquer problem, LeetCode 395.
In the first-round HM interview, the kind interviewer let me through.
Second round:
-
A 30-minute ML system-design interview. The interviewer asked me to design a shop-logistics system. Grocery-store products spoil easily, so I had to design an algorithm that determines a range for the amount of inventory to order.
-
A 30-minute ML domain-knowledge interview. The interviewer said I did not know anything. Honestly, it did help me find gaps, because there was a lot I had not prepared: tokenization methods, dense versus sparse retrieval, what an RNN can do besides time series, and explaining a Gaussian distribution. It felt like there was a question bank. Unfortunately, I could not answer them.
-
A 60-minute coding interview. The interviewer simply disappeared. Emailing HR did not help either, and I do not know whether I had already failed the previous two interviews.
Although the experience was just average, it was genuinely useful for finding gaps in my knowledge. I received an automatic rejection email the next day, which was not surprising.
Discussion
Loading comments…