As a child, I used to try and count the number of stars in the sky, absolutely mesmerised by the twinkling dots. From the long nights on the roof looking above in fascination started the admiration for astronomy.

As a junior in high school, my older cousin brother introduced me to Machine Learning. Seems weird to say that, but that’s how things panned out. He was working towards his PhD at the time in swarm robotics and gave me a primer of artificial intelligence. Ever since then, I started to read up more on the subject. On reaching college, I pursued a few projects to get a better feeling of academia and the research environment.

During my sophomore year at BITS, GSoC was recommended to me by many of my seniors as a nice way to work and capitalize the summers. I glanced through the participating organizations and OpenAstronomy caught my attention. Specifically, the CTLearn sub-organization seemed like a match made in heaven, using Machine Learning algorithms for gamma-ray data.

I quickly introduced myself to the community and went through the codebase. The community was very warm and welcoming. When I was a bit comfortable with the code, a few tasks were put forward, to assess our skills and understanding.

Navigating through the code and documentation while performing the tasks was a very enthralling experience. I thoroughly enjoyed it and learned a lot in those few weeks. Soon, I created the proposal, made some edits and submitted the application.

Damian Lillard hitting the game-winner over Paul George, sending the Thunder home. Game 5, Playoffs 2019. Approximately encapsulates how I felt reading the acceptance letter.

When the acceptance came through I was on cloud nine. Nothing beats that feeling of accomplishment and the urge to perform. I soon contacted my mentor and we discussed out a rough timeline of the project. I was also assigned a few tasks, in order to better align the project’s direction.

While I slogged of on this assignment, one error kept bugging me. Initially, I was hesitant to ask my mentor for guidance this early on, but I finally gave way and pinged him. He immediately realised that it was in fact a bug in the latest build. So I fixed the bug, which resulted in making the code a bit faster as a consequence. That first merged PR felt nice in the heart, and I know I'm in for an amazing ride ahead.