CareerCup

Tell me about a time when you not just met your goals but exceeded the expectations

Write code to return true if a cycle exist in a graph.

Input to function is a 2D array where every array was of size 2 and contained connecting vertices.

For above question, asked to do coding in C Lang.
He asked me what data structure to be used for storing employee records.
Then about the uses of arrays, hashmaps.
Which has more advantages if we want to search for a record with employee name?

Assume there are employee records with all required fields. like name, emp_id, manager_name etc..

Write a SQL query to print the 2 employee names whose manager is same.

What is process synchronization, why do we need volatile keyword if we need to use mutex locks

Volatile keywork --> what is the use of it, why compiler has to do optimizations, in which cases u have used volatile.

Given a text file - find out all the unique words and display the frequency of each word in descending order.

Given a input string 'Hello World' and a input reference String, give count of each character (from refString) in the given string
Eg inputString: "Hello World Please", refString: "Help"
Output: H: 1, e: 3, l: 4, p: 1

ELK related questions especially on optimizing logging

Design a Nearby shops features for online map service provider.

Design a system to find top 10 sold products for an online website at any point of time for last 15 minutes.

Which data structure & algorithm would be the best to design such kind of systems?

You are given a directed cyclic graph represented by an adjacency matrix. The graph has at least one terminal node (i.e. the node with no outgoing edges).

Each edge of the graph is assigned a positive integer representing a probability of taking this edge. E.g. if you have 3 outgoing edges with numbers 3, 2, and 4, this means that the prob. of taking these edges are: 3/9, 2/9 and 4/9, respectively.

You need to find the probability of reaching each terminal node from the starting node 0.

Example:
adjacency matrix 5x5:

m = {{0 1 0 0 1},
        {0 0 3 2 0},
        {4 0 0 1 0},
        {0 0 0 0 0},
        {0 0 0 0 0}}

so we have two terminal nodes here: 3 and 4
we can take the following paths:
0 -> 1 -> 3 = probability: 1/2*2/5 = 1/5
0 -> 1 -> 2 -> 3 = probability: 1/2*3/5*1/5 = 3/50
0 -> 1 -> 2 -> 0 -> 1 -> 3 ... and so on

or to the node 4:
0 -> 4: probability: 1/2
0 -> 1 -> 2 -> 0 -> 4: probability 1/2*3/5*4/5*1/2 = 3/25

and so on, summing up probabilities of all possible paths we get:
total probability to reach node 3: 13/38
total probability to reach node 4: 25/38

design an ip blocking system

Given an n-ary tree and some queries for the tree, in every query you’ll be given a node you are supposed to print preorder traversal of the subtree rooted at that node.

You have an array of numbers. You have to give the range in which each number is the maximum element. For Example, If array is 1, 5, 4, 3, 6 The output would be
1 [1, 1]
5 [1, 4]
4 [3, 4]
3 [4, 4]
6 [1, 5]

Given an array of billion of numbers. Billions of queries are generated with parameters as starting and an ending index. Both these indices lie within that array. Find the maximum number between these two indices in less than O(N)

Multiply two numbers without using * and only be using bitwise operations

Display nodes of a tree in level order using DFS

Implement a deque using stacks

Given the entire dictionary of English words, what data structure will you use to efficiently store and match the custom regex string like "D*sk", where * represents any single alphabet, and return the list of matched words?

Given a skewed tree, an insect is sitting at the root of the tree at t = 0min, every minute insect steps down in the tree, find the probability of the insect being at any node at t = infinity. Once I came up with a solution various other complexities has been added to the problem such as: What if the tree is binary tree (written code for this) What if three is n-ary What if it is now a directed acyclic graph Handle cases that there can be more than one entry point There can be more that one way to reach a node

Given a binary tree of numbers and a search number has given, find out first occurence of that number and smallest distance from root node. if you have given k search numbers find their occurence and nearest from root node in a single walk.

Given a string of numbers put commas so that it become readable like million trillion thousands. eg 1010503 ===> 1,010,503

Given an array of 0s and 1s. find maximum no of consecutive 1s. If you have given chance to flip a bit to 1 such that it maximises the consecutive 1s. find out that flipped bit and after flipping that bit maximum no of consecutive 1s. Above question but you have options to flip k bits.

Design a system which will keep track of product and its inventory count. The service will expose two api incrCnt(int prodId, int cnt) and decrCnt(int prodId, int cnt). Which db would you use ? How will you handle hot products ?

Design a system in which read a file which has data and operation to be performed give a line by line. Ex a=5, next line b= 10, next line a*b. This design extended to support float, doubles, boolean, vector and complex numbers. Like if the file has a=5+i8, then how you handle such scenarios. How you will store and process data.

Design content ingestion system

You are given 2 arrays: one representing the time people arrive at a door and other representing the direction they want to go(in or out) You have to find at what time each person will use the door provided no 2 people can use the door at the same time. Constraints: the door starts with ‘in’ position, in case of a conflict(2 or more people trying to use the door at the same time), the direction previously used holds precedence. If there is no conflict whoever comes first uses the door. Also if no one uses the door, it reverts back to the starting ‘in’ position. Should be linear time complexity.

A startup website has a lot of real-time traffic . I want to see the real-time view (refreshed every 1 min) of top 20 users by hit count within last 10 mins. Full distributed system, I have to resolve all the concurrency issues.

I was asked this during my onsite google interview but was unable to come up with an optimization for it. Here is the question:

There's a list of (x,y) points and a method getCircle with the following signature:

/**
* Given three points returns a circle (Radius, and center) such that all three points lie in its circumference
* or it returns null if no such circle is possible.
*/
Circle getCircle(point1, point2, point3);
getCircle method is already implemented and given to you as a black box. The problems asks you to find the Circle with most points in its perimeter.

The obvious answer is to get all possible triplets of points and find all possible circles and keep track of which one appears most often O(n^3) . Any ideas on how to further optimize this?