First Semester of Ph.D. in Review and some Reflections

Research Block at IIIT Delhi, taken from here.

I have recently finished the grading work of my first course as a TA, and done with all my coursework. So, I find this as an excellent opportunity to write my first blog post. The end of the semester has always been about reflecting on how I could have made it a better experience. This time it’s a bit different because I am now a Ph.D. scholar with a subconscious worry about my work and Covid-19 lockdown.

During the last semester of my undergrad, I joined as an intern at MIDAS Lab and continued as a Research Assistant. In December, I decided to take up a Ph.D. in the same lab because I always wanted to get one, and IIIT Delhi is one of the best places in the country to get a Ph.D.

My time as an RA gave me an idea about the course loads at the institute, which helped me narrow down my choice for courses. I wanted to work on my mathematical maturity early on and still wanted to take courses related to my research, so I decided to take Convex Optimization and Statistical Machine Learning, which turned out to be fun and useful, and it was nice taking those. For my TAship, since it was my first time, I consulted my advisor and ended up being a TA for his course, Multimedia Computing, and Applications.

Being an RA before joining as a Ph.D. student allowed me to continue working on the same projects as a seed problem for my Ph.D. research. It wasn’t easy to maintain a momentum of research while diving into coursework head-on. This hard-to-maintain balance is the first lesson I take for further semesters, my research flow should not break at any costs while I make sure that I enjoy the courses I enroll in, which will require smarter scheduling and time management on my part.

Due to Covid-19, courses moved online, which took out some of the fun in taking these classes. Nonetheless, one of the most significant impacts of this lockdown is irregular sleep schedules and very fluctuating productivity. Some of my colleagues feel the same way.

Coming to project management, this is where I slogged a bit while writing code and managing experiments. I am working on my project alone, so there is no second member to enforce quality constraints on my codebase. My project involved data scraping, a lot of preprocessing, and a lot of parallel training of models. The last two things are still going on, and the codebase has grown and evolved in unforeseen ways. The lesson that I learned here is to ensure strict guidelines to handle how my code changes and structure out everything created over the past week. Time spent there eventually gets spent later once when you sit down and clean your code. I realize this skill will take time to develop, and thanks to an increasing emphasis on reproducibility in my field, some very well written codebases are getting released, and just going through them teaches me a lot. On the bright side, I experienced exhausting all inodes on a drive of around 10TB. This issue is something a few of us will face, but keep an eye out in case you are working with videos.

On the bright side, I finally set up a blog. This lockdown gave me some time to work on some math concepts (more on that in later blog posts), and have some good time with my family. It made me realize something I took for granted, wandering around the campus, sitting in the garden, and the library. I hope we return to college soon.