Amazon Interview Experience | Set 183 (SDE – New Grad Position)

Following is my experience. I hope it helps people out there, as I got help from other posts here before my interview.

Duration: 4 hours

interviewer #1: 8 am – 9 am

Q1) tell me your interest
-> Big Data
Q2) what’s latest thing you follow in it
-> Apache Spark
Q3) What is it and explain
-> told the standard MapRed and Hadoop vs Spark
Q4) what more features
-> told graph processing and MLLIB
Q5) Actual question starts here, given the binary tree, how will you serialize and deserialize it to the file?
->
answer1: Serialize => do in-order traversal and pre-order traversal and write it to a file
               Deserialize => build the binary tree using pre-order and in-order traversal
     he was not satisfied, gave me hint of no need to do complex in-order, pre-order thing

answer2: BFS clicked me, told him to do BFS and write to a file level wise
             But he found a bug in deserialize method, I changed the serialize method for null values to speial character in file

he was okay with it, and asked me to write a code for it
asked the complexity
told him, was not convinced wholly, then finally told him, was still not
left the room with think over it 🙁

interviewer #2: 9am – 10 am
Q1) bang on target, given 2 timeframes, check if they overlap
-> s1, e1, s2, e2
   ep1 = epoch(s1),
   ep2
   ep3
   ep4

   if ep1 <= ep3 <= e2 OR ep3<= ep1 <= ep4:      return T    return F Q2) LCA in doubly binary tree !!! -> I frowned, what is doubly binary tree!!
   it’s just every node has access to the parent

but you need to find the LCA and you do not have access to root of tree!
some brainstorm, and I got it
what is time complexity:
worst case: O(n^2) in skewed tree
average case: was not sure, gave me hint, and I told him, was okay

told him the 1st way, was convinced but hinted me the better way
Got his hint, and gave the answer, was satisfied

Q3) given 2 strings, how would you find if they are anagrams of each other
-> solution 1: 2 hash ( was okay but asked the better way)

hinted me that no of characters are limited
solution 2: told him the way of count sort, i.e. use 2 count sort arrays
but he said no need of other array, you can do in single array

told him the way, was satisfied

Interviewer #3:
– started with his long intro
– asked mine, and then stopped me as I was talking at length
– asked me amour my search project
– asked me the basic design
– how did you deal with data when you cannot fit in memory
– how did you do intersection of posting lists
– what is time complexity

then asked me to design the chase game
stumbled like hell, i guess this is going to be my rejection factor

Interviewer #4: (manager) behavioral round
– gave his long intro
– asked about myself
– when did you feel that something could have been done in better way after delivering it in project
– when did you perform best way, as in what did you come up with that other could not do
– when did you go out of the way i.e. against your managers, or higher authority and deliver the best according to you

– asked the system architecture question: design the TINYURL system in terms of server architecture i.e. HA, scalability and request processing

– was satisfied, and left with good note.

All Practice Problems for Amazon !