Amazon AUTA SDE 1 Interview Update
I am currently working in a Big 4 with (1.3+.06)yoe. Facing "depraisals", facing PIP for crossing Manager. Explaining Manager that more people doesn't mean faster work and handling Excel sheets for survival.
I have been grinding Leetcode since 4th Year. I have applied to Amazon for 70+ times in last 3 years.
I applied to Amazon AUTA and completed their test in March. There were 2 questions.
Q1. You have a set of blocks of dimension 2×1. You can either place it vertically, so that it takes 1 unit of horizontal space, or horizontally, so that it takes 2 units of horizontal space, and we can stack 2 blocks, one over another. You can colour each block in 3 colour's, such that no 2 blocks have same colour, horizontally or vertically. How many colour combos are possible.
Q2. You are given an array and integer. You can select a subarray and multiply it with the integer or divide with it. Calculate the greatest subarray sum you can get by performing 1 of the operations.
Completed both of them. Recieved Interview Call in May, to Hyderabad.
Round 1:
DSA and LP
Q1. Binary Search in Rotated Sorted Matrix. (EXACT)
Q2. Modified LCA, where each node can have any number of children, and we need to find out the distance between the 2 nodes.
Was asked to define the class structure and constructor for the same.
Verdict: Cleared. The interviewers even praised my communication and explanation skills(Thanks to Big 4 to make an Engineer, who just wants to be a Developer, to handle clients every day for a living)
Round 2:
TRAIN WRECK!!!!!
The Interviewer was an Senior SDE, with 5+ yoe in Amazon. He deep dived into LLD!!!
I HAD NO IDEA ABOUT LLD, SOLID AND WHATSOEVER.
Was given a problem on Designing a package like UNIX Find. Where I need to write classes find files according to their extensions, and their size.
I first started by saying, that I am assuming that the files are stored in cloud, since that's what I have been working for past 1.9 years. Making an index DB with string name, double size, string location (Address to body in ADLS)
His reaction, why cloud storage, why are you making the problem difficult. He then asked me to define all the classes.
I first started to create classes for CRUD.
His reply, Why are you focusing on CRUD. I am only interested in Retrieval.
I then started with creation of overloaded functions, where if number and charecter for comparison is entered, it will query for every entry according to the operator and threshold. And if string is entered, it will check if the string starts with ".", then it will query all the file names, and do string matching, later updating the schema by adding another column as file type.
He then started asking about what will happen for multiple queries.
I again went on with function overloading, where one will be the usual, another will be taking an added parameter of a vector, from the previous query.
He dismissed it entirely, saying my code doesn't have scope of Scaling/Changing.
My last effort was to say that we can declare an array of size = entire db, and increment the counter for each cell for every query it matches.
Again he dismissed by saying, I am more focused on the solution than the structure.
Then he jumped to LP, and asked about a time I did something outside my responsibilities.
I started with an experience where I sacrificed my WFH and went to their office in my hometown to tutor a senior on the projects tech stacks (To someone, who has 5 times my experience, getting paid thrice as much as mine, reported my work as his own, and had me thrown out of the team😅)
He bluntly said that he is not interested about how I helped out at teammate, and wants an answer that is oriented to me using a different techstack.
I shared another experience, which again he contradicted. I ended up explaining to him what a demerger is and how the original table is NOT a parent table.
Verdict: REJECTED
Well, not everyone has got a happy story to share, ig!! Back to handling Excel Sheets...
😅😅😅