I applied through a recruiter. I interviewed at NVIDIA
Interview
NOTE: interview process was during the pandemic shutdown, so fully remote
-one exploratory interview, about an hour long
-two rounds of technical interviews, each broken up into three 45 minute one-on-one segments with different individuals
-HR was responsive and friendly when I reached out to them with questions
Interview questions [1]
Question 1
-Questions about deep neural network design and training
-Other general more general ML and data science questions
-Language/framework specific questions
The overall experience was decent. The first-round interview was reasonable: the interviewer asked detailed questions about distributed training and went through my past project experience. However, the first round explicitly told me the coding portion would not be LeetCode-style questions, but the second round only consisted of LeetCode problems, which I had never practiced. This mismatch between expectations and the actual interview was the main issue.
Interview questions [1]
Question 1
Can you explain the details of how distributed training works, and walk me through your past projects?
One round with the hiring manager on the basics and advanced topics of deep learning, covering the latest developments of transformer architecture as well, parallel training methods, and optimization techniques.
It was good, but interviews were hard
Some of the interviewers didn't seem interested (didn't even turn on his video)
They asked me two leetcode hards in 40 mins which I thought was crazy