Back to The Software and Analytics Guide
Introduction
Hey everyone! I am Suyash Kamalakar, a fourth-year Chemical Engineering undergrad. I did my summer internship under the Autonomous Driving profile at Robert Bosch (ECE). I come from Kolhapur, Maharashtra. I like to sketch, write and play cricket in my spare time. Also, I love reading about personal finance and innovations in tech!
For my third-year internship, I had decided to focus on Analytics and Software roles as my resume was inclined toward them. I was certain about not going to core and non-tech roles. I had a couple of projects in Machine Learning, but no development experience. In the winter break before the 4th semester, I took an SoC in Competitive Coding, just out of curiosity. That was my first introduction to Data Structures and Algorithms. During the 4th sem, I solved some questions but I was not consistent at it.
Selection Procedure
When the tests for the companies started, I saw most companies ask Dynamic Programming and Graph questions in the initial round. I had practiced some DP questions, so I was able to solve, sometimes partially, a few of the problems, but never enough to get me shortlisted. I had made it to the waitlists of some companies but understood that more rigorous practice was required. Thus, I started solving questions especially from DP.
Also, I was of the belief that there won't be any good offers after September (which turned out to be false). So parallelly I started approaching recruiters on LinkedIn and applying on other career sites. Bosch came around the end of September with 2 profiles, Autonomous Driving (AI) and Cyber security. After reading the details of the first profile, I saw it resonating with my interests and applied for the same. They announced a resume shortlist and luckily my name was there. It was mainly due to a project in object detection that I made during the summer break.
Coming to the interview process, it was a single round, mostly focused on resume projects and some ML questions. We discussed my project in depth, then the interviewer asked which ML algorithm I am most comfortable with and later asked me to tell me the math behind it with some follow-up questions. To check my coding ability, he also asked an easy string-based question and allowed me to code in Python. The interviewer was very friendly and helped me whenever he found me stuck. The interview ended with some common HR questions. I was elated to see my name in the final selections, finally.
About the Project & Company
Initially, we were introduced to the project the team was working on and given some basic reading materials to get started. Then, specific materials for object detection were shared and we were divided into 3 groups. My 3-membered group was on domain adaptation, with each of us given a subtopic. We were working independently. My project was Domain Adaptation in Object Detection using Image-Image Translation.
It was a completely new field for me, so I was very excited to work in it. I will summarize my work very briefly. After learning about the working of GANs and CycleGANs, I did a literature survey to find the latest approach for image translation. I applied that approach on 3 European traffic sign datasets and then performed object detection using Faster-RCNN. I was able to reduce the cross-domain effect using this technique. Finally, I did some experiments and validated the effectiveness of the approach. The duration of the internship was around 80 days, which gave us enough time to get familiar with the Bosch cluster and solve the errors arising during the internship but at the same time, we could not get any holidays during the summer break.
I love the automobile industry and I keep reading and watching about the innovations in this field. Autonomous Driving in itself is very fascinating and contributing to a project focused on this field was truly a great opportunity for me. I learned a lot of new things during this internship like standard practices in developing an end-end deep learning pipeline, data preprocessing, and choosing the suitable model with parameters etc.
Talking about other aspects, Bosch is one of the very few companies that offer internships in Computer Vision. The office is situated in Adugodi, Bangalore. The office timings were 10 am to 5 pm and weekends were off.
The mentors and managers in our team were extremely skilled in their fields and were very supportive. There were also extra sessions conducted on the topics of our choice and the sessions were full of information, from the basics to advanced concepts. Mentors were always ready to help with anything that we asked for. The atmosphere in the office was very positive and overall the experience was enriching.