The process took 2 days. I interviewed at Amazon (Seattle, WA) in Sep 2011
Interview
The big thing wasn't exact coding skills; it was ALL problem solving and worthing with a client. IE, you are the engineer, the interviewer is the client, how do you design and lay out what they need their software to do etc. The big thing was how do you build a system; I focused on making cheap, easily maintainable systems, and that was impressive to them.
I also point out some flaws with the assumptions of the questions (this isn't how all users act, etc) and that helped out a lot as well.
Even when I didn't fully understand the problem, I kept asking and figuring, trying to learn what it is that they wanted, and how I could build it. You should always talk out your thought processes, to show them they are happening, and to get them to talk as well! TALK TO GET THEM TO TALK!!
Interviewed for silicon team. Have only been asked about the domain specific knowledge in 1st round and system design in 2nd round and C coding in 3rd round.
The interviews were 50 mins each.
First round with hr screening - 2 leetcode questions then hr manager screening then the loop which consists of 4 interviews each an hour long. The 4 interview questions they asked where three medium leetcode questions. And one system design interview question about how to shadow deploy a test software to millions of users.
The phone screen went longer than expected, focusing heavily on implementation details. The interviewer really grilled me on my approach to a Least Recently Used (LRU) cache, asking how I'd combine a hashmap with a doubly linked list. I felt well-prepared since I had gone through system design examples on PracHub, which made me comfortable discussing eviction policies. The later rounds included more technical questions and behavioral interviews, but in the end, I received an offer, though I ultimately decided to decline. Overall, I’d say the process was average, with solid questions.
Interview questions [1]
Question 1
Design and implement a Least Recently Used (LRU) cache supporting get(key) and put(key, value) in O(1) average time. Walk through combining a hashmap with a doubly linked list, eviction policy when capacity is exceeded, and how you'd extend it to handle thread-safe concurrent access.