Akshit Bhalla

I aim to do meaningful work in the field of data science. My interests broadly include Machine Learning, Statistical Analysis, and Computer Vision.

I spent my summer observing research involving Reinforcement Learning at the Indian Institute of Science (IISc), where I was fortunate to find a great mentor and peers who built a quadruped. I chose this particular concept because I just thought it was a really cool project. I learned that to really go beyond the hype of Machine Learning and to create something that wasn't an imitation required digging deep and experimenting with the math.

Reinforcement Learning turned out to be one of the most interesting things I had ever dealt with. It is very much an "in development" technology and no one has guidelines on how to go about it. Building the hardware for the project was relatively straightforward, most parts were 3D-printed, but integrating the software to get it working was a nightmare.

A large portion of time went into background research on the locomotion and bio-mechanics of different animals and the types of gait each used. Horses and dogs both use 4 legs to move around but are fundamentally different in the way they do so. A cheetah leaps while a deer might trot. Analyzing the data to draw meaningful conclusions, for training the model, was the most important part of the process in hindsight.
It was hard work, but watching the creation walk awkwardly was so much more fulfilling than I had been expecting. The next thing I'd like to do is teaching a quadruped to dance to music.

I realized that Data Science was something I could take up as my career and still be interested in 10 years down the line. I intend to become a post-doctoral researcher in this field. I spend my time reading and trying to implement research papers, an altogether tough endeavor to do on my own.

I started NeuralDot (https://neuraldot.github.io), a student research group in my college that discusses interesting papers and implements them. I hope to publish a research paper of my own with the findings I make along the way.

The world is pivoting towards a future wherein these technologies are increasingly important. I am excited about the potential that deep learning holds and am passionate about contributing responsibly to this field.