Status: Experience as Software engineer and ML engineer, B.Tech in Mechanical Engg (from Tier 1 college)
Position: Machine Learning Engineer at Google
Location: Bangalore, India
Date: 5th December, 2019
I have been contacted by a google recruiter for their bangalore team. During introduction call she asked me time complexities of few sorting algos and few questions about machine learning basics(confusion matrix, train/validation set split).
Round 1:
Type: Telephonic
Duration : 45 mins
Interviewer briefed me about google and interview process for first 5 mins and asked few questions about my experience.
Then next 15 mins he asked multiple questions about machine learning: precision/recall, how to avoid underfitting/overfitting, what is regularization and dropout.
For next 20 mins he asked me to implement k-means clustering algorithm in any of my favourite language on shared google doc.
Last 5 min he asked me if I have any questions for him.
Recruiter told me that they still need more information before calling me F2F hence fixed one more telephonic round.
Round 2 :
Type: Telephonic
5 mins: Discussion about process/ google and my exprience
next 15 mins: Few questions about programming languages (python vs java), then basics about machine learning: gradient descent, overfitting/underfitting, k-fold validation etc.
Then he provided me following coding question on shared google doc:-
https://leetcode.com/problems/missing-ranges/
Then recruiter asked me to come F2F in their bangalore office for 5 more interviews:
Onsite Interview 1:
Design question :
You are given user_id, timestamp, search_query and list of links clicked by user while doing google search. Design a machine learning system which predicts which links to show based on search query.
Onsite Interview 2
Design Question:
- Design machine learning system which shows spelling and grammer errors and also suggest suitable correction to them while user is typing on google doc.
- Given the output of above model, code the logic in any preferred language (I choose python) which can be used to provide real time correction and predictions to user while he is typing.
Lunch break with existing google employee
Onsite Interview 3
Behavioral :
Describe situations and how did you handled them when:-
-
Someone else took credit for your work
-
Disagreement with collegeue
-
Disagreement with management
-
How do you manage team
-
Project you are proud of
Onsite Interview 4
Coding round:
Given a list, find out k elements which produces max sum while you can only take continuous elements from either end.
For eg : [5,4, 3, 10, 1, 1000, 2]
for k = 2, elements would be 1000 and 2, hence max sum = 1002
for k = 3 elements would be 5, 1000 and 2, hence max sum = 1007
for k = 4 elements would be 5,4,1000,2, hence max sum = 1011
Interviewer told me the question verbally and couln't explain hence I have spent 15 mins trying to understand what she wants me to code. I have still completed the question within 45 mins but got negative feedback.
Onsite Interview 5
Design round: Detect advertisement boxes within youtube videos using machine learning.
Coding round: Given above machine learning model, Write code in any language which returns frame range/timestamp range and x,y position of the all advertisement boxes in a specific video.