I applied through a recruiter. I interviewed at Meta (Boston, MA) in Oct 2020
Interview
I had three technical interviews. The first one didn't go very well. The second and third went great.
Then I had three team match interviews. The first was not a good match. The second and third were good matches. One of the matched teams made me an offer, and the other went ahead with a different candidate.
Interview questions [1]
Question 1
These questions were asked:
1. Implementing an arithmetic operator without using it or equivalent ones.
2. Subarray sum with high efficiency.
3. Postfix expression evaluation.
Describe the complete process of training a VL reward model. They ask you how the dataset is curated, which model to use, and how to get the reward from a generative model
Interview questions [1]
Question 1
Describe the complete process of training a VL reward model. They ask you how the dataset is curated, which model to use, and how to get the reward from a generative model
The interview process is smooth. I got two round interview, the first is a research interivew and the second is a technical interview. Research interview is related to my past research and the technical interview is related with practical problem in the research field.
Interview questions [1]
Question 1
Describe how the attention mechanism works inthe transformer
It was a three-interview process: one coding interview and two technical interviews. The coding problems were not that difficult, but the technical interview was very hard. I have two technical interviews as I got a neutral result from the first technical review.
Interview questions [1]
Question 1
Coding
Q1. How to pool the 2D conv net when the context length is limited.
Q2. Find an error in the given code.