I applied online. The process took 7 weeks. I interviewed at Amazon in Nov 2021
Interview
The interview for PhD students is fairly straightforward. You give 2-3 interviews which include technical and behavioural rounds. It can take around 6 - 8. weeks to complete the interview process.
Interview questions [1]
Question 1
Few recommendations,
- Be ready for typical behavioural questions. Preparation beforehand is a must.
- LC easy-medium questions.
- Be ready to answer questions related to your past/ongoing projects. Sometimes, the interviewer can go deep into the technical details of the project.
- Explore Amazon research beforehand and figure out where your research work fits best and if possible mention this in the interview.
- Last but not least, prepare a few questions you would like to ask the interviewer in the end (there will be dedicated time for that)
Overall, it's a great experience and you get to learn a lot. Best of luck!
2 interviews in total. 1 ML round and 1 coding round. The interviews were supposed to be spread across 5 days but due to scheduling issues (interviewer not being available), they actually took 15 days to be scheduled. Results were also delayed due to the holiday season.
Interview questions [1]
Question 1
ML round
- asked questions about overfitting, model design, precision, recall, F1 scores
Coding round
- design algorithm to implement byte pair encoding
I applied for generic position ,
recruiter recommended Alexa team
I had really bad experience in my phone interview . Interviewer was from Alexa team
I have shared some of the questions that were asked and our conversation
question after the interview by me. can you tell what kind of projects are you working on. interviewer gave some vauge answer and asked to look at their publications
what is the difference between rnn and lstm. answer by me lstm has a extra forget Gate to implement rnn can't remember long context and theoritically lstms can remember all the relevant history even if it's very far away.
he said it's a vauge reply
then interview asked about demnishing gradients which I explained , and he asked lstm vs rnn which one handle the dimnishing gragient problem better. for which I replied lstm
to which he said you should have told this in lstm vs rnn
at this point I was really disappointed with his knowledge , tone and attitude . but what to do he was the interviewer