Status: 3+ years Post-doc
Date: May 3, 2020

Technical phone screen (1 hour):

• Classical ML questions
• Behavioral Questions
• Coding question: Graph-based, medium difficulty

Thoughts: I gave an average performance in fact. But I felt i did the behavioral part well and I got invitation for the second round a few days later.

I started following some online ML lectures (MIT: Patrick Winston) and also started Leetcode. :)

Technical phone screen 2 (1 hour):

• Classical ML-based solutions for a Toy problem
• Coding question: Simple sorting

Thoughts: I aced this one. The online lectures really helped and the coding part was easy.

I received the invitation for the final rounds. It was a virtual interview involving 6 rounds and I did it in two days (3 each per day). I had roughly two weeks to prepare for the final rounds.

Virtual Onsite Interviews

1. Round 1: Technical Presentation (1 hr)
   Presentation on my recent work. (45 mins presentation and 15 mins qns)
2. Round 2: Technical Interview (1hr)
   • Deep Learning based questions
   • ML questions related to the job
   • 30 mins of behavioral questions
3. Round 3: Technical Interview (1hr) (I think this one was the bar raiser)
   • Deep Learning + Classical ML based questions
   • 30 mins of behavioral questions
4. Round 4: With Hiring Manager (1hr)
   • 30 mins of behavioral questions
   • Questions about my presentation, coding practices
5. Round 5: Coding (1hr)
   • 30 mins of behavioral questions
   • Coding question: Graph-based
6. Round 6: Coding 2 (1hr)
   • 30 mins of behavioral questions
   • Coding question: String-based with some system design

Thoughts: The interview process was quite smooth. The interviewers will give hints if you are stuck. I felt I had too many behavioral questions (half of my whole interview), but I think Amazon gives a lot of importance to them.

Preparation: As a post-doc I already had hands-on experience on many ML and deep learning techniques. But I have not taken any formal courses in ML. So for the interview, I did several online courses for preparation. Introduction to algorithms (MIT), Deep learning course from CMU (Bhiksha Raj), ... and a lot of leetcode. In fact, the final coding round was something I saw in leetcode and I aced it. For the behavioral part, I prepared several stories in the STAR format and I practiced them. I did not know half of my interview would be behavioral kind of questions, so I reused some stories, twisted them a little bit to fit to the question. I watched videos from Dan Croiter, it was very useful.

I received the job offer in a week and I accepted it after some negotiation. :)

Hope this helps and I wish you good luck!