About me
What do I do?
Research: I'm currently focusing on pain and aversive (safe) learning in context of decision-making but am interested in all sorts of interesting behaviour shown by humans and other animals. I employ emerging technologies such as virtual reality into my experiments and ideas from machine learning to build computational models of animal cognition. I then try to translate these computational models to algorithmic ones, working towards neuro-inspired robot learning. I truly believe in a symbiotic relationship between natural and artificial intelligence and am optimistic about how research in one field can guide the progress in the other. My background is in electronics engineering (at BITS Goa) and scalable deep learning (at Nvidia).
In my free time, I enjoy fiddling around creating digital art, listening to music, reading books (usually non-fiction but ocassionally fiction), watching anime and reading manga and swimming or playing cricket. I seem to have taken a recent interest in learning about value investing and vtubing.
Interests:
Computational Neuroscience
Safe AI and Robot Learning
Virtual Reality
Philosophy
Compounding
Education
DPhil Clinical Neurosciences
University of Oxford
B.E. Electronics and Communications Engineering
BITS Pilani, Goa Campus
The rest of this page is a self-indulgent piece of writing with no target audience in mind.
So I've decided to pen down a mini-auto-biography/timeline for context and because why not, this might be all that's left of me one day.
1999: Born, very grateful.
2012: Started getting better and formal training in mathematics and sciences from a first principles approach
2014: Created and published my first video game
2017: Started my undergraduate in engineering in Goa. Proceeded to spend three to four of the best years of my life at Goa.
2021: Started my DPhil in Oxford. I enjoy my work, most of it doesn't feel like work. I enjoy the intellectual freedom.
2022: Have made a VR experiment using Unity (game engine). Life has indeed come around a full circle.
If there's anything else you'd want to know and think would be better off being on this page, kindly let me know. What follows now is a brain-dump of my thoughts about me (since you clicked on this page to know more about me).
Why do I do the stuff I do?
Time and again, people I know, I enjoy spending time with and people who are doing very well in their life at different stages of life ask me, why do I research pain? Some of them are puzzled to see someone with quantitative background not pursuing an engineering (computer science, electronics or financial) or a consultancy job. While some see the pressing needs to improve pain management globally, my research seems far too focused on basic science questions, at glance atleast. It'd be noble to say my main motivation is to strive towards better healthcare and develop neurotech therapeutics but, to be honest, it's probably my secondary motivation. So why do I research pain?
Because I love learning, especially learning how people learn and really, how any animal learns. In words of Naval, "The source of wisdom is pain". The most unforgettable lessons learned are the ones in moments of pain and suffering. Societies are build over pain, some of the strongest bonds over a lifetime are forged in times of suffering. And it's also worthy to speculate the significant role of pain in evolution itself, or as to why we are the we are today.
Pain is not just a sensation, a feeling or an inference. It's a motivator for action, it drives learning, it makes feel alive and also makes sure we stay alive. Reward for long has been known to drive learning, how a school of thought has come to develop to almost (unfortunately) equate dopamine to reward prediction error. Much of the pain side of the story is almost forgotten and not received the attention it deserves, furthermore pain is special because treating it as just negative reward in existing frameworks, is far inadequate to capture the reality of pain and it's interesting properties such as endogenous modulation, subjective experience, single-shot learning, empathy and learning from others, reflexes, injury and persistence of chronic pain, to name a few.
I think, we are the most alive in moments of pain, it detaches you form all sorts of distractions (that we call life) and really demands and focuses your attention towards itself like nothing else. When asked and given a choice, most people would choose to eradicate all pain and suffering from the world we inhabit and rightly so - much of it needs to be gone but not all of it. Some of it is exceedingly necessary for what we are today and where we as a society of living beings are headed towards.
In a broader context, why do I do science? Over years, I find myself trying to figure out and really understand invariant principles guiding this world. Usually most of them tend to be time-invariant laws i.e. they stand the test of time and generalize pretty well. It was never really the games that interested me, to be honest, I'm not very good at most of them, but it's usually the rules of the game that I want to know. The best games are ones which show a rich span of behaviours using the simplest and as little rules as possible (for example, Go). We don't necessarily have to but much of the world we live in could be thought of as game (or a simulation, in one of three possibilites of Bostrom's argument), and much of science as I understand is figuring out the rules of the game. Much of artificial intelligence as I understand is making something that can figure out the rules and learn to play the game rather than fiddle around with poorly provided instruction manual. I can assume, a few of the readers would be exasperated to see me conflate rules and laws (especially cricket-lovers), but bear in mind, the game is an anology here.
This idea of finding time-invariant laws seems to have infected all of my other excursions, one of them being compounding. It's rather unsurprising that many scientists, including Einstein have realized the power of compounding at some point in their lives, yet it uneases me to see only a fraction of few scientists I know act to take benefit of it. Value investing, as I am realizing, is only an intelligent way to take advantage of it and make the law work in your favour.
What do I struggle with?
I struggle with letting go of this notion of optimality. It can be seen in my daily life and also my research. Reinforcement learning has been a field I picked up early on and it assumes having a reward function or a cost function and learning an optimal policy which optimizes it. This certainly has been proven to be very useful in explain how humans and animals might learn and it certainly captures a lot of rich behaviours in mechanistic manner with decent neural correlations. But then is life really a quest of optimization? If yes, it'd be just a matter of finding out your inclinations, preferences and what you value in life and then clearly optimizing for it from the get go. Much of this problem can be seen in sort of inverse RL, which starts by finding out what sort of function does this agent or animal must be optimizing to achieve the behaviour it is showing currently. Sort of trying to distil and reduce anything and everything to cost functions to optimize. Have we built a hammer that we have gotten so used to, that everything starts looking like a nail to us? It certainly did to me for a while, and I struggle with unlearning it. I struggle with not always trying to find the optimal way out in my daily life. I struggle with riding the flows of life, as points in attractor spaces. Everything I do doesn't need to be optimal, sometimes it's okay to just exist.
What drives me?
Acknowledgements from a handful of people I admire. Most probably has roots in my child psychology. People I admire might change through life, ought to change, but this again seems to be a time-invariant driving force in my life.
My life philosophy
Make life simple.
I thank M. Prakash for saying these words to a class of students one day, when I was 14 years old. It was of course in a very different context, he was trying to teach us mathematical problem solving techniques for math olympiads. But this razor or a thumbrule generalizes well out of context and has stuck with me since then.
A few things I've come to believe in -
No great work was ever done without solitude
The best thing you can offer to the world is you. (not your research or your ideas)