Sanidhya Anand - Universitat Wien

Hello there, I’m an undergraduate student in the Energy Science and Engineering Department - yes, a DD student writing about his intern, so you have an idea of what to expect. By the time you read this, I should have been accepted for an IDDDP in CMInDS with an MTech in AI & Data Science, with these two being what my professional interests have converged on. I am currently interning as a Student Research Assistant under Prof. Nils Kriege at Universität Wien in Vienna, Austria, and writing this on the 81st day of my 89-day Europe trip work-visa permitted stay in the Schengen region.

Pre Intern Season - The Allure of Corporate Moniz

I did not start the internship season wanting to do research. I was fresh out of a very eye-opening internship with an Indian think-tank; it had made me realise that I did not want to work in my department’s non-core area (the profile was energy-consulting but in a less consult environment, iykwim). I had been dabbling in Machine Learning through my CS minor and found it *chef's kiss*; I was pursuing a project in the time-series domain with a professor in collaboration with a Germany-based start-up in the summer (this eventually collapsed in a marvellous fashion, so not as impressive as it sounds).

Still, I thought it would be hard for me to level with my peers who had been familiar with ML since their first year. I knew no seniors interested in core ML and had no interest in diversifying into Computer Vision yet (if you have even heard about ML, you know how popular CV is). Come July, I was left with practising a lot of DSA, revising a ton of ML concepts and the only thing I had finished on time was my resume, which was lacklustre by the standards of a good tech resume.

And yes, I panicked a little. But this is where my DD status came in; the thought that almost no company would open for me calmed me down - you can’t fail at an opportunity you never got *finger guns*. I was mainly looking for well-paid ML RnD roles (I was more of a sellout back then), and through days of looking at companies, I realised that only 3-4 such positions would be opening up for me in the first two months, and they would have sky-high levels of competition (damn you, EE department kids). Still, I diverted my focus on practising whatever coding I could (all hail Hackerrank) and sat through hours of Kylian Weinberger's golden lecture offerings at Cornell, which I did not end up finishing, but the effort counts.

Intern Season (*Insert scary music*)

For the underprepared (most of us, really), the entire intern season is like going to the gym for the first time. You’re extremely nervous and afraid of being the only inexperienced one there, and it doesn’t help that the guys around you look like beefcakes benching 220 pounds casually. I was giving tests for whatever software company opened up with even a remote mention of ML, and any parallel aspirations I had of getting into non-core companies went down the drain with their pre-final year restrictions. Getting through core software tests was tough enough as it is with my make-shift DSA skills, and when the ML RnD companies started shortlisting me for interviews, they were all looking for people with CV in their basket.

By the end of September, I had given seven interviews and been rejected in all of them. IAFs opening for profiles I wanted became scant, and almost all of my friends had internships by then. I realised the anxiety accompanying intern season was taking a toll on me. I had a 53-credit semester going on and desperately needed a break - I completely pulled out from the process. I had picked up an RnD with the same professor from my previous project, on Graph Machine Learning, and diverted my focus to my academics so that I would have some semblance of progress in ML.

Intern season, in a nutshell

At this time, the senior I was working with on my RnD asked about my intern season, and that one half-hour talk did wonders. She encouraged me to apply to professors internationally, introduced me to DAAD (Graph ML is big daddy in German Universities) and assuaged my insecurities about my branch negatively affecting my chances - professors really don’t care about branches if they think you’re good enough at what you do. My DD branch provided the flexibility to work in a 4th-year internship if I wanted to, so a 3rd-year research intern sounded like a good idea. I was getting more and more habituated to reading literature in ML through the RnD, and eventually decided to start applying to European and American professors because I now had a somewhat educated idea about their work. I created professor databases, started monitoring whether they were looking to accept students and getting up at odd times to mail them - LDAP mail scheduling is not a thing.

Come November, and I had ZERO replies on the day of the DAAD deadline. Still, I shrugged it off and diversified to regular apping to non-German professors, even putting a Day 1 intern friend to unpaid work on my database. Eventually, he found Prof. Kriege, and I got called for an interview. He had a smaller team and was actively looking for students to work under him, and I got the sense that I would be held to higher expectations. And surely enough, he made me write an entire research proposal in the months of February and March as opposed to just an SoP to secure funding under his program.

Austria, but with a picture of a rainy Budapest, Hungary

I’m here for the maximum time my temporary work visa would allow me to be, and I will say I have enjoyed my intern. I meet my professor weekly to discuss problems and results; the ideation was more or less finished by my first three weeks as I had already worked on the proposal. Talking about my research, I’m basically trying to figure out if two nodes in a graph are similar using their neighbourhoods. Once that was implemented, I extended the (secret) idea to different types of graphs and continuous node feature embeddings extracted using neural methods. The method is simple enough, and I had to implement the entire codebase from scratch while also verifying the effectiveness on datasets that were either available or I created. Very often, mental breakthroughs came from staring at drawings like the one below. Pretty much standard research :D

I had never travelled internationally, let alone doing so alone. Living in Vienna for the last three months has been a very unusual yet warming experience; I’ve learnt how to cook (barely), take care of household chores, develop my own work-life balance and become comfortable with living independently. Hopefully, if you plan to do an intern in Europe too, you’ll have a ton of friends with the same idea - if not, then you’ll probably make friends with the IITB people there anyway.

What I’m taking out of this internship (and what you should take out of this blog)

I have been convinced, well enough, that I want to pursue a career in Machine Learning, and I’m actively leveraging the confidence and experience this intern has given me to apply to other projects and eventually corporate RnD internships. This was not what I planned to do when I was in your place (if you’re a third-year student at IITB), but it all worked out in the end. And trust me, so will it with you. Intern season can be a behemoth to tackle, and I think my unique take on it is that it’s okay to take a break from the race once in a while, if that means you’ll be returning with a stronger resolve. And if you want to ditch the race altogether, there are countless opportunities out there to apply to, with the only condition that you look for them. All the best, and feel free to look for me if you have any questions or even just want to talk! :)