Fractal Analytics DataEngineer Interview Questions 

By Karthik Kondpak

Fractal Analytics  rounds

==========================

For some There will be online MCQ round also 

1) Technical Round 1

2) Technical Round 2 

3) Managerial Round 

Sometimes 2nd and 3rd round is combined as One

Technical Round One:-

====================

1) Tell me whole architecture of your project?

2. Tell me about spark architecture?

3. How spark runs in standalone mode?

4. Tell me how spark divides the program in different Jobs, stages and tasks?

5. How spark decides where to launch the executor in cluster?

6. What are the roles and responsibility of driver in spark  yarn Architecture?

7. On what basis yarn resource manager decides to allocate resources to 

spark?

8. How spark allocate memory to executors?

9. What is Namenode? What is secondary namenode and what it does?

10.Why spark is so much famous? In what case will you prefer hive instead of 

spark?

11. Tell me one full pipeline that you have build and tell me its Data size, cluster 

capacity and execution time in detail?

12. If I have 500 GB data then in order to process it what will be my ideal cluster 

configuration? Explain me in detail.

13.You have an employee table (Id, name, address, pin). Write sql to get top 5 

employee name having pincode ending with 1. And then write the same in 

pyspark using dataframe api?

14.You have an employee table (empname, department, salary). Write sql to get 

top 5 employees in each department based on salary and also write the 

same in pyspark using dataframe api.?

15.Write a program to reverse a string without using reverse function. Write your 

own implementation?

16.What is your current cluster configuration?

17.What are types of tables we have in hive? Where table information gets 

stored in hive and when we can use external table?

18.What is the difference between client and cluster deployment mode?

19. If your spark programs fails then can you gracefully handle it?

20. If everyday you are getting some sales file in sales folder and one day by 

mistake someone placed invoice file in sales directly then how you spark 

application will behave? How can you handle this situation?

21. How are you taking the existing code and working on top of that? How is your 

current ci/cd pipeline works?

22.Which IDE are you using to do your development work?

23.What is difference between rank and dense rank function in sql?

24.What types of join have you used?

Technical Round 2

==========================

1)How are you handling data skewness?

 2. What is the role of zookeeper in hadoop cluster? 

3. Do you have any experience in K8s?

 4. What automation tools have you used in your big data project? 

5. What happens if any of the Datanode goes down? How will you handle this one in production?

 6. How spark decides which join strategy to use?

 7. If any spark application fails then what is your approach to troubleshoot that?

 8. What algorithm resource manager uses to schedule the spark jobs?

 9. How can you minimise the data shuffle in join in spark? 

10.What is the maximum size of table that we can broadcast in case of broadcast join?

 11. How shuffle hash join different than sort merge join? 

12. Did you work on any sql optimisation? 

13. Let's say you have an array = [3,34,4,12,5,2,9]. Write a python program to find all the possible sub array's from this array whose sum is equal to 9