Basic Information

  • Name: Abhijat Bharadwaj
  • Email: 210020002@iitb.ac.in
  • Curriculum: Dual Degree
  • Year of Study: 5th
  • Company / University: Qualcomm
  • Role: Computer Vision
  • Duration: 25 Summers
  • Mode: On-site
  • Location: Hyderabad

Targeted Domain

I was mainly targeting fields like Computer Vision, Audio/Speech Processing, etc., which are joint domains of ML/AI and signal processing. I had some prior experience in software development, so I tried for software roles as well.

How I Got This Internship

Short answer, networking and cold mailing. I got to know about the opportunity from an alumnus and reached out to them regarding it. Apart from this, I sent around 30-40 applications for various roles at different companies.

Application Timeline:

  1. I mailed my application in December ‘24
  2. Had technical interviews in January ‘25
  3. Heard back from HR in February ‘25
  4. Received offer letter in April ‘25

Preparation Resources

For preparation, I referred to the following materials: 1. Course Materials: Digital Signal Processing (EE338), Image Processing (EE610), Intro to ML (EE769), Computer Vision (CS763, EE702), Wavelets (EE678), 2. Striver’s Sheet and Geeks For Geeks for DSA and Interview Questions, and 3. The projects on my Resume. But as things always go, some of it was useful, some was not.

Tests and Interviews

Most of my interview was spent on my resume. I was asked to elaborate on some of my projects (especially my DDP). Discussions were about the algorithms and reasonings, project workflows and why I made certain design choices in the projects. I was also questioned on fundamentals of ML, Computer Vision, Image Processing. The programming questions were easy and could be solved with basic DSA or ML programming background.

Handling Rejections

I knew a lot of my early rejections were because companies either preferred a 6-month intern or candidates with a better CPI. For cold mails, it is anyways tough to expect replies, let alone a positive one. Also, I was a bit selective about the roles I wanted to apply to in my 4th year, which it made it even more difficult to expect openings. So I was already prepared to not expect much out of most of the applications.

I think what works best is that you need to keep networking, whether actively or not, reaching out to seniors, alums, profs, etc. Applications for interns are already crowded and having a good network of reliable people helps navigating it a lot.

Work / Project

I worked on exploring and designing “Deep Learning Techniques for Global Motion Estimation” (a sub-topic in computer vision). The project was at its early stages when I started and my work was about finding different ways to approach the problem. I worked on established CNN models for homography estimation and SoTA 3D scene reconstruction models. The work was part ML engineering and part research oriented.

Organizational Culture

My team had a really friendly and welcoming culture. Most of my interactions with them was informal. There was a lot of freedom and flexibility in one’s working style, and people genuinely seemed passionate about their work. Qualcomm promotes regular team outings and team bonding events, which my team planned during my internship period. Even people outside my team were really approachable and extremely helpful.

Networking

Yup, the people are really approachable, irrespective of their positions and titles. Qualcomm also encourages team outings, intern outings, internal hackathons, etc. which makes networking a lot more easier.

Overall Experience

10/10. Amazing work and brilliant people.

Would You Recommend?

Yes, definitely! In the field of computer vision, there are a lot of practical and interesting problems which are either left unsolved, or partially solved. The R&D work of academia and industry are working in really different ways and the work in this field is often really intellectually stimulating, especially in leading companies like Qualcomm.