class undergradAtPrecog(IIITD):
def __init(self)__:
How I got to know about Precog is quite dramatic (call it fate? hahaha). It had been a really long day (Intro to Engineering Design project demo, IIRC) and I was walking past GB Pant towards the metro station. To give a rough idea about how far back I’m talking about: Phase II construction of the college was yet to begin, attendance was never recorded in any lecture, and Ubers were allowed to enter college premises.
As I was walking, absolutely dreading the hour long ride back home, a car stopped by and offered me a ride till the metro station. And I begin an extremely foolish (and lazy) kid, got into the car without giving it a second thought. However, the rest of the story doesn’t pan out like a Bollywood kidnapping thriller. To the contrary, I met two of the most fantastic people I know, Megha and Rohan! It was during this ride down to Hauz Khas metro (yep, they saved a lot of the metro anguish that day), that I got to know about Precog, PK, and most importantly that I don’t need to have taken any advanced courses in order to join Precog (many professors ask you to take their course before joining in).
Unfortunately, I wasn’t able to apply straight away (mostly since I was spending any time I got those days writing code for FLINT). However, I did realise that there is a lot more to computer science than Byld and Foobar (which I was completely obsessed with during that period). So as soon as I wrapped up my GSoC project, I shot PK an email asking whether they’re taking any students.
My application process was very straightforward. After sending in my CV and SoP, I was called directly for a brief chat with The (now Dr.) Dewan and Anu. Both of them have mentored and assisted me for extended periods with my projects, and I’ve learnt much more than just how to write research abstracts from them (including PTK’s art of cracking the best puns ever, and Anu’s art of subtle and effective sarcasm). Of course, I did destroy multiple servers in the process, but it was always forgiven within a week with restored access 🙂 After the first round, PTK and Anu recommended me to PK, who took the final screening.
All in all, it was a really smooth and professional process, much more to the point than the majority of the job interviews I’ve been taking recently. What still surprises me, is the fact that I’ve never come across another lab member who wasn’t given a task during the interview process. Guess it became mandatory after I clogged up a couple of servers hahaha?
My initial time at the lab seemed to be somewhat jinxed I’d say. The first two projects I worked on between Jan 2016 and April 2016 got scrapped in succession due to some technical issues. Both involved some analytics on data curated by an external agency. At this point it felt like a “local minima”, and I started wondering whether I was doing something wrong. When I meekly pointed it out, thankfully PK intervened and explained how it was not at all an issue on my end, and I should be expressing my thoughts more frequently rather than overthinking. That’s one thing I’ve always kept in mind, and tried to improve upon. Yes, folks, communication is the key.!
The summer of sixteen nights
During Summer ’16, PK gave me the opportunity to go spend some time at IIIT Hyderabad. They were hosting IASNLP over two weeks (16 nights :P) and had invited some people from Precog. He explained that we would like to do some work with code switched social network text, and gave me a free hand to Indira, Mallika and I to work on whatever we like.
IASNLP proved to be an extremely important stepping stone for me, as it was my first introduction to computational linguistics and machine learning (something I am working on till date!) Upon returning, we finalised our target to be a set of low level tools (LIDF, POS tagger, NER, Sentiment) for English – Hindi code switched text. Although I didn’t realise it back then, the problem was not as trivial as I thought. In fact, I ended up working on it for almost a complete year (taking it up as my BTech Project eventually), and finally published it at ACL this year (link and link).
Lab members and interns, summer 0f 2016. Missing a lot of people here
Why Precog?
Being a part of Precog is much larger than just completing your thesis/ project. When you’re a member of Precog, you’re a member of Precog. Weekly group updates keep you up to date with what’s going around in the lab. Regular paper reading sessions keep you updated with the cutting edge in areas other people are working in (I’d admit I wasn’t able to take full advantage of this, something I deeply regret). If I had to cherry-pick five top reasons I’m gonna miss being here
- I got access to whatever resources I required to work. Anshuman, Divam and I constantly used to fight over GPU servers. PK noticed and ordered a separate one for each project 😀
- Members are extremely helpful. If you ever need to get a code or paper draft review, all you need to do is shoot an email on the mailing list. And it is not just work. There were countless times I’ve pestered Anu/PTK/Srishti/NS and even PK to review my SoP for various applications!
- There’s no need to be diplomatic. When it comes to work, people expect you to speak freely what’s on your mind. Even if it hurts sentiments.
- Just like the IIITD culture, there is no concept of senior / junior / “call me sir/ma’am” amongst members. I think we inherit this really well from our college. Anu kept a special jar where you had to put in 10 bucks every time you called her ma’am :’)
- We don’t just work together, we party together too, and a lot! Apart from regular PK treats at Barbeque Nation and sometimes even his apartment (how many of you can boast of having had dinner in the faculty residence :P), often we head out to GK after a tiring week (PTK always seemed to be fixated with The Beer Cafe for some reason).
You never walk alone. Winter 2017 semester, I took a break from college. On the eve of Precog’s birthday, members called me up via Skype for the celebrations 😀
To say Precog has played an instrumental role in shaping who I am today would be a massive understatement. I’ve spent more than half my undergrad life associated this lab, its members and its wonderful wonderful machines (JARVIS, FRIDAY and DEXTER). I have failed to produce results multiple times, missed multiple paper deadlines and annoyed a lot of members. However each time, I’ve learnt something really important, which allowed me to succeed in the end. And oh boy, when I finally got a paper accept (two in a row in fact, ACL SRW and SocialNLP @ ACL!), it was a beautiful closure.
PK and I after my (almost) convocation in August this year
What I’m grateful and indebted for, is the fact that I was allowed to make mistakes and learn from them. An essential piece of PK “gyan” (knowledge, for the non-Hindi speakers) which I’ve valued increasingly over the years, is that “working hard is good, working smart is better”. Choosing your battles is as equally important as working hard to win your battles. To succeed, I think it’s essential to surround yourself with people that believe in you. My journey with Precog was certainly not easy, it had its fair share of ups and down. However, it comprised of everything that was required for me to get out of my comfort zone and succeed.