{"id":6246,"date":"2026-08-05T09:35:40","date_gmt":"2026-08-05T09:35:40","guid":{"rendered":"https:\/\/precog.iiit.ac.in\/blog\/?p=6246"},"modified":"2026-08-05T09:35:41","modified_gmt":"2026-08-05T09:35:41","slug":"more-than-a-lab-my-experience-at-precog","status":"publish","type":"post","link":"https:\/\/precog.iiit.ac.in\/blog\/?p=6246","title":{"rendered":"More than a lab: My Experience at Precog"},"content":{"rendered":"\n

Hello! I am Vaishnavi Shivkumar, a recent graduate of the UG2k22 ECE batch, who had the wonderful opportunity to work with Precog-at-IIITH<\/a>. I\u2019ll start this blog by saying that the experience of working with Precog has been nothing short of transformative. The conversations I have had, the work that I have iterated over, the people I have met \u2013 all of it affected me in a very distinct way, which has given me learnings for a lifetime. But I shall tell you that story slowly.<\/p>\n\n\n\n

This blog is going to be about my experience at Precog, and why joining this place is a canon event for me.<\/p>\n\n\n\n

Part 1: At the beginning<\/h3>\n\n\n\n

Sometime around January 2024, I stepped into an \u201cIntro to Precog\u201d session where Ponnurangam Kumaraguru (PK)<\/a> told us what Precog does. Knowing less about research and lesser about Precog, I walked into that room with no expectations. The room was packed with students \u2013 like me \u2013 who were here because Precog had a lot of buzz around it.<\/p>\n\n\n\n

The first thing I realised is the type of quiet ambitiousness that PK carried into that room. Existing Precog seniors sat amongst us, and one by one told us about the projects that they worked on. PK said one thing, which sticks by me as a summarizing learning of the entire lab.<\/p>\n\n\n\n

It\u2019s a place where ordinary people do extraordinary things.<\/em><\/p>\n\n\n\n

At the time, I was a doe-eyed student with a new amazement with NLP, and a fire to prove that I could do something that I picked up from ground zero. That environment of determined ambition inspired me. The promise of freedom was invigorating, and the confidence and jovial seriousness that PK pitched his lab with, made me curious to see what it all was about.<\/p>\n\n\n\n

When we were finally asked to give our statement of purpose, and given a task to do \u2013 I knew that I wanted to do it, just because it seemed very<\/em> fun.<\/p>\n\n\n\n

With due time, my doe-eyed excitement got me into the lab, and I instantly got a taste of what happens when you put a bunch of high agency, high ambition people in a place that they value. Our Precog seniors organized a long introductory session (which spanned for 2 weeks) to tell us everything they learned \u2013 from tech tools to research practices. Those tutorials served to be extremely useful; and was a setting to learn about the vast world of research.<\/p>\n\n\n\n

The immediate thing I realized is that everyone is happy to help you here. Conversations after tutorial sessions, or fixing a coding bug, or having a discussion about their ideas \u2013 they were open to help. I used to look at these seniors with awe, but from the onset they treated me like a person who could actually<\/em> do anything.<\/p>\n\n\n\n

That is something that you will see as a pattern in Precog; softly reinforced by the way PK interacts with us. He strongly believed that as a lab we could do anything if we put our head to it, and that \u2013  subtly, and slowly made me believe it too. He ensured to set the bar high, whether it is credibility of work, thoroughness of writing and experiments, or goals and conferences \u2013 he expanded what we believed was possible, and did it with a nonchalance that made you feel silly for not believing you could do it in the first place.<\/p>\n\n\n\n

In the beginning, the possibilities seem endless. Luckily, in Precog; the possibilities are always endless.<\/p>\n\n\n\n

Part 2: What it means to be part of something bigger<\/h3>\n\n\n\n

A key way to how Precog worked was, as mentioned earlier, was helping each other. This took shape in the form of shepherding (reading and tracking progress of other\u2019s papers), Weekly Updates, frequent lab discussions and more.<\/p>\n\n\n\n

I remember, for a while in my third year, I did not get why we had to meet so much<\/em>. Only when I came to my fourth year, did it start making sense \u2013 that is, when I had an idea of my own. That was when the continuous intense feedback, the intuition checks, the design philosophy discussions, thinking out loud about the bigger picture, presenting presenting presenting .. all of it made sense. There was a time when me and my co-authors stayed in the lab for 19 hours a day for a week; and the way the lab & PK showed up for us \u2013 constantly asking about progress, giving advice and support, even getting food and coffee! \u2013 it all made me realize that damn! <\/em>I was part of something genuinely bigger than me.<\/p>\n\n\n\n

This was further established in Weekly meetings. PK often opened the floor for all of us to pitch in, on where we want to take the lab, how we want it to operate. That kind of clarity, and equality in operation was really refreshing. We were able to be honest about our fears and our concerns along with being able to pitch new directions. Whether a student said little or vehemently disagreed with a direction \u2013 they were allowed to say it. PK insisted that there were no wrong answers.<\/p>\n\n\n\n

This way, we all built the lab. I remember PK often telling us that he has a lot of faith in the process of how Precog operates \u2013 that is, the efforts, the clarity, the structure. And that is a learning that I take with me through the rest of my life.<\/p>\n\n\n\n

Overall, being part of something bigger, forces you to be the best version of yourself. And I can only say that I tried :><\/p>\n\n\n\n

Part 3: What it means to collaborate and iterate<\/h3>\n\n\n\n

One thing I learnt from all the technical work I did at Precog was structured execution and design thinking. We spent a lot of time running experiments \u2013 and through the process of trial and error (and a truckload of advice from senior Precoggers!) I slowly learnt how to be procedural about my ideas.<\/p>\n\n\n\n

Initially, it feels like every idea you stumble across is your magnum opus (probably have had over 100 such delusions xD), but breaking it down step by step, designing a schema of why the idea is worth its salt, creating experiments that work towards or against the concept \u2013 these are all things I learnt from Precog. PK always encouraged us to get our hands dirty \u2013 write an abstract, run some experiments, pitch an idea to the rest of the lab.<\/p>\n\n\n\n

This way we truly learnt to market <\/em>and find a story around our idea. This process: 1, gets you to be excited about what you are doing which works as fuel to keep pushing and 2, gets you to find all the gaps in the story that lingers in your head.<\/p>\n\n\n\n

We were always pushed to work hard and fail fast! Better to be ripped apart by your fellow lab mates rather than the reviewers \ud83d\ude00<\/p>\n\n\n\n

Precog, hence, forced me to show up. Show up not only for the lab \u2013 but also for myself. The environment pushes clarity with a speed \u2013 which in turn makes you a better version of yourself. Of course, with a lot of people by your side.<\/p>\n\n\n\n

My first project in Precog was working on CodeMixing Model Merging alongside Prashant, Swarang and Anisha. At the time, I was new to running large scale experiments \u2013 which didn’t seem to deter my collaborators. I would often ask a lot of questions, ranging from how to use ada to why my bash script wasn\u2019t working (#!\/bin\/bash vs # !\/bin\/bash \ud83d\ude4f). I was incredibly lucky to work with Prashant, who was extremely patient, and kind with me as I figured out the tech stack needed. Seeing his genuine excitement about our topic, and continual iteration motivated me a lot, and inspired similar excitement in me. We had to run a lot of experiments, and I learnt how to navigate that process without stumbling. Working with a team of motivated, kind, funny and intelligent people who really cared was such a blessing!<\/p>\n\n\n\n

My second big project was working with Srija and George on Fairness in Federated learning. It taught me a whirlwind of things \u2013 from design thinking, to ground up coding but most importantly it gave me incredibly good friends. I still cherish the amount of fun we all had in the lab spending equal (if not more) time talking and working. We all took up different experiments, often challenging and  ragebaiting (friendly) each other to continue to iterate. Each bug in the code would take another iteration on Srija\u2019s ipad, which would lead to trials and trials and trials. We would start the evening with one idea, and end the night with another one \u2013 ending with the sun rising as we walked back to our hostels discussing everything under the sun.<\/p>\n\n\n\n

The thing no one tells you when you are ambitious is how hard it can be to get back up when you fail. But the thing that Precog gave me, was a group of friends and collaborators who were just as stubborn to do things right. And that makes a world of a difference. Speaking of which\u2026.<\/p>\n\n\n\n

Part 4: What failure teaches you<\/h3>\n\n\n\n

I have spent quite some time thinking about how I phrase this blog, and this section is one which has stayed constant. One of the most transformative aspects of my time at Precog, is the way it forced me to confront failure and taught me to move ahead from it. You see, researchers, small or big, fail a lot. We have ideas that we need to stress test, encounter knowledge gaps that we need to cross and deal with bugs that break our entire code. There is a struggle to the process of becoming good at anything, and that struggle often goes unstated in blogs like this. There are ideas that we cannot articulate perfectly, where we spend nights and nights reading literature so that we have the language of even placing<\/em> a question.<\/p>\n\n\n\n

Naturally hence, every person in Precog has faced their fair share of failures. At the risk of getting vulnerable, I am truly grateful for all the big and small ways people in the lab uplifted me from failure. Whether it was stupid jokes, or serious conversations; the faith that the lab showed me in action, is the reason I continue to believe in myself.<\/p>\n\n\n\n

Let\u2019s start with placing the failures. At Precog, we gave our 100% to the projects. That is the nature of work that demands a lot out of you. Through paper rejects, through a code bug that broke our entire idea a week before a deadline, through getting good ideas and finding that it’s already a Zhang et al in ICLR 20(current_year)-1, we all were forced to carry failure as a friend in the process. The good thing about Precog though (inadvertently advocated by PK), is that it sees you for the efforts you put in. I remember being scared that I would be reprimanded for failing; just to get a \u201cgood job\u201d from PK. When we spent weeks trying to break our hypothesis, with repeated experiments, and new and newer directions, PK silently noticed it all, and supported us. When it all ultimately didn\u2019t give us a result, he acknowledged that he saw how hard we worked on it.<\/p>\n\n\n\n

In technical institutions, such humanity sticks by you. When nothing I was experimenting on worked, I still remember juniors in my lab genuinely rooting for me and listening as I discussed my ideas out loud and collaborators encouraging and advising me till the last hour. Those are things which cannot be replicated without kind people, in a community that thrives when everyone wins.<\/p>\n\n\n\n

Precog was one such community, holding both quiet and visible spaces for these losses. We were encouraged by PK to take breaks when things got too hard, to take care of ourselves, with an assurance that he believes in us, and that we would come back stronger. The linkedin posts and selfies of Prof PK may show a little of his personality, but they do not even scratch the surface of how much such small words from an advisor and a lab mean to someone who faces rejection.<\/p>\n\n\n\n

There are other things that failure teaches you. It teaches you to cut out the noise, and relentlessly look for clarity, with intention, calmness and precision. It reminds you that you cannot do things by yourself, and makes you shameless in asking for help. It shows you the importance of fighting, and reframing your hypothesis into a story that is coherent and without holes. It makes you respect the people who are the most honest to you, either with feedback or support. It teaches you to push past your own self doubt with determination. And finally, it is a true test of how much something matters to you.<\/p>\n\n\n\n

I still remember the small ways in which the lab and our collaborators helped us keep going: a senior sitting down with you an hour before a deadline to dismantle and rebuild your abstract; watching a seasoned researcher return after one or two rejections, determined to make the work better; and seeing how far our own minds could stretch for an idea we truly believed in. These moments taught me that resilience is rarely an individual achievement. It is built through examples, conversations, acts of generosity, and the quiet knowledge that, even when the work fails, you do not have to face that failure alone.<\/p>\n\n\n\n

Part 5: What freedom gives you<\/h3>\n\n\n\n

Finally, the crux of the selling point of Precog \u2013 one which you may have heard about time and time again. Freedom.<\/p>\n\n\n\n

I walked into this lab with no idea what I wanted to do, and with no strong intuition about ML, and no real reason to do it,except that it was fun. I did not fully know what it entails to be a researcher.<\/p>\n\n\n\n

Precog showed me that I can be a person who can rely on my ideas, and take the arduous steps needed to build it up. It gave me a community where I could be free in discussion and deliberation. It gave me the privilege of knowing a professor who rooted for me through a myriad of failures. Now I know what it’s like to have a group of super ambitious people around me, and how I myself respond to it. It taught me about my own resilience, and how I could be wrong, and wrong, and wrong again\u2026 and still keep moving. It expanded my idea of freedom, and forced me to fight for my ideas, and fight to make them into a reality.<\/p>\n\n\n\n

Part 6: The Acknowledgement section is important!<\/h3>\n\n\n\n

If it isn\u2019t clear enough already, the design of the lab, and the way PK runs the place is front and centre to the reason Precog remains so important to me. So this is a personalized acknowledgement section to my very own camera ready blog xD<\/p>\n\n\n\n

To Ponnurangam Kumaraguru (PK)<\/a> : thank you for always bringing humanity into the lab \u2013 for being serious when required, tough when required, but always making it clear that you cared first and foremost about our growth. And, of course, for building a place that I now find so difficult to leave.<\/p>\n\n\n\n

To my close friends: Karuna K Chandra<\/a>, for giving me a friendship I know will last a lifetime; George Rahul<\/a> , for ensuring that I did far less work and had far more fun; Debangan Mishra<\/a>, for our cherishable talks and for being my fellow lab snoozer; Krishak Aneja<\/a> , for the endless conversations, strange humour, and food recommendations; Srija Mukhopadhyay<\/a>, for being so wonderfully similar to me and for all the stupid jokes; Sumit Kumar<\/a> for the constant excitement; Vamshi Krishna Bonagiri<\/a> , for treating me like a younger sister; and Hari Shankar<\/a> for the friendship both inside & outside the lab.<\/p>\n\n\n\n

To Prashant Kodali<\/a> , Sreeram Vennam<\/a> , Shashwat Singh<\/a> and Akshit Sinha<\/a>: thank you for helping me stress-test my ideas and telling me when they did \u2014 or very much did not make sense.<\/p>\n\n\n\n

To Aditya Mishra<\/a> , Manas Mittal<\/a> , Arihant Rastogi<\/a> , Sujay Belsare<\/a> , Vishal Rao<\/a> , Harini, Varshita Kolipaka<\/a> and everyone else who filled the lab with conversations (serious & unserious), tomfoolery and laughter, thanks for keeping me sane!<\/p>\n\n\n\n

And to everyone I had the privilege of collaborating with \u2014 including Professor Saurav, Professor Sheetal, Professor Monojit to labmates \u2013 Swarang, Likhit, and many others, thanks for the patience and guidance!<\/p>\n\n\n\n

Part 7: Pictures, Contributions & More!<\/h3>\n\n\n\n