Returning To Academia

Tech industry internship; A practical detour from academia.

The past 9 months have been a very interesting time for me.

I started an internship over the summer, which lasted for nearly a year. Overall it was an incredible experience where I learned so much. I was able to spend a lot of time discovering real problems that are unsolved in natural language processing, and in general for machine learning. These are different problems from what shows up in the big-name papers at any research conference.

I dealt with problems like having messy, unlabeled data. Data so difficult to label that even a team of experts can’t agree and how to train a classifier on these types of data (turns out there are some methods, but they don’t show up in papers because many papers use clean benchmarking datasets).

I dealt with problems like not having nearly enough data. For some current SOTA models, this is a big issue because they are so data-hungry. There are also solutions for that.

With all of these difficulties though, came an opportunity to design clever solutions. Solutions that are admittedly not perfect, but certainly unique and unheard of in scientific literature (patents soon to be filed).

Over the course of my internship I have not only gained knowledge about the field of NLP, but also refined my own interests (somewhat). My interests are still very broad, but for the purpose of PhD research, I have been able to narrow down my focus. An industry internship has been hugely helpful in narrowing this down due to a gained understanding of what is of practical use in the field, and what some of the remaining open problems are.

Now, I have completed a second leg of my long PhD journey. Although I’m not technically done with my internship as of today, I have already mentally looked back and reflected on what it has done for me. I’ve contemplated what I have been able to accomplish for the company, and I am thankful that it has led me to where I am today, prepared to make contributions to the scientific community.

Next step in this journey is to settle down in my new lab, get to know my colleagues, get to know my professor, and extend the limits of knowledge.