Internship Experience at IIT Bombay

Hey everyone!

I’m Pranav Hari Panicker, a B.Tech 4th Year student in the Electrical Engineering Department. This past summer, I had a great opportunity to work on a research project as an intern at IIT Bombay. I had a great deal of interest in the area of Machine Learning & Deep Learning, which gave me the incentive to discuss with some of the professors at IITB online and to try and work under their guidance. Subsequently, one of the professors selected me through the official project application process to work under him and his team (Prof. Surya Durbha - Center of Studies for Resources Engineering). It was an enriching experience, to say the least. As it aligned with my interests in the field of deep learning, I thoroughly enjoyed every aspect of it.

As a first-time research intern, I absolutely had no idea how it would pan out. I also had certain qualms about research internships owing to my lack of work in any prior projects as such. Since I was never that interested in or exposed to research before, I had minimal expectations from it. On the contrary, I was quite happy with how things turned out. Discussions with my mentor resulted in many learnings and takeaways that were incredibly helpful. Also, the other PhD scholars in the department labs really helped me with their valuable insights & suggestions.

Broadly and vaguely speaking, the main area of focus for my project was the applications of deep learning to geospatial data. Initially, I had to go through certain tutorials and website documentation to get the hang of what exactly the project demanded from me. With the help of the PhD scholars already working under the Prof, I got a basic idea of what they were trying to do i.e., create a model using the commonly-used libraries of scikit-learn and TensorFlow for mapping soil moisture data in a 2D spatial surface. I was tasked with aiding them in further enhancing the said model with different & seemingly more efficient techniques.

With regard to this, I began working on a new framework altogether, known as GPflow, to implement the same model more effectively. Constant support from the PhD guides during my frequent visits to the labs of the CSRE department at IITB (since I had a hybrid internship) helped me throughout this process of using GPflow to analyze 2D soil moisture data. I also spent a few days testing out and experimenting with a few of the devices in the geo-technological lab, which added a lovely and casual touch to the internship. To be frank, the offline aspect of the internship was like no other, as it provided an enjoyable and full-fledged learning experience.

I had also been using other DL techniques like SpaceGAN, a form of generative adversarial networks (GAN), to create new data points in 2D space to help make up for the limited data points that were provided. Once again, it was back to the basics for someone like me, who had no idea about it. Learning and understanding these new and never-seen-before-or-heard-of technologies was the most exciting part of being involved in this research study.

My implementation of these newer methods turned out to be quite satisfactory to my Prof and the rest of the team. Though there’s still quite a lot of work left for them to continue, I was able to come to an understanding that my work had ended up setting some basic groundwork, and it was beneficial to my own learning curve as well. One of the main things I’m looking forward to is having regular post-internship sessions with the guides from the dept (be it casual or work-oriented), just to keep in touch and have some fun discussions with them.

Ultimately, I’d just like to impress upon exactly how enriching and insightful this experience was to me. For someone not quite accustomed to research and projects in general, it seemed to me that I was in a whole new dimension of experiences during every single eventful discussion/meeting, which taught me something better almost every day.

There probably might be a few students who may seem hesitant in looking forward to accepting research internships, but let me tell you that the learnings you take away end up being just too incredible & memorable. If you’re just too damn interested in a certain field within your branch, and you’ve got such a research opportunity at hand, best believe you wouldn’t regret choosing such an experience whatsoever.