Write the first damn draft yourself

I’ve said this before and I’ll say it again:

Using LLMs-written content professionally will spoil your career, especially if you’re in early stages. Everyone worth their salt that I know is hugely aversive to reading slop and despite your smart attempts, they can sense signatures of AI writing (even if they can’t articulate what gave it away).

When you delegate writing, you delegate thinking and that delegation takes away from you the very thing you need for your future career advancement: careful thinking. ...  Read the entire post →

Unsolicited advice to the Indian govt on frontier models

What can 🇮🇳 do to get ahead

Here’s some unsolicited advice to the government on what to do to help boost India’s sovereign model story.

1. Mandate domestic models for all non-critical functions

Government needs to be a guaranteed buyer for homegrown models. There are tons of applications that don’t require SOTA performance. (To be clear, for mission-critical use-cases, using anything other than SOTA would be a shortsighted decision).

China got ahead because of their domestic demand (Anthropic and OpenAI models don’t work there).

2. Release massive distillations of chain-of-thought of SOTA Chinese models on diverse prompts (and post-process to remove China bias)

Inference is easy and cheap. So the government should simply create a national public repository of SOTA reasoning chains in order to prevent each model creator doing it separately. Centralize and absorb the cost. ...  Read the entire post →

[2025 review] Life is a holiday!

Ok, this was a phenomenal year for me. The vibe of the year is cleanly captured by what I had scribbled on a whiteboard on Jan 3rd, 2025.

To my amazement and giddiness, the feeling still holds up as the year wraps up. Let’s see what all kept this spirit alive.

Brief interlude: the yearly review has become a sort of an annual tradition for me. In case you’re interested, you can check out previous years’ reviews too: 2024, 2023, 2022, 2021, 2020 and the entire decade before that.

🫡 Exited Wingify

I had started Wingify in 2010 (when I was 22) and in Jan 2025 (when I was 37) exited my stake to Everstone, a private equity firm. As you can imagine, Wingify consumed the entirety of my early youth. It taught me about startups, business, leadership, human psychology and put me in contact with an incredible set of people who I’m proud to have worked with (especially my co-founder Sparsh and CTO Ankit Jain). ...  Read the entire post →

Notes on doing science and research

I know it’s been awfully quiet here. I’ve not posted for a while.

But I’ve been writing prolifically somewhere else. Earlier this year, I started an AI lab called Lossfunk and that is what has kept me (happily) busy. We have a newsletter there called Lossfunk Letters where I write very often, so if you’re interested in my thoughts, please follow that publication.

The initial posts there are about me trying to build a mental model for what is that thing we call scientific research. I explored this topic via a series of posts: ...  Read the entire post →

Becoming a Republic

The following is the essay I had submitted as part of the one week summer course I took at the Oxford University, where we read Plato’s Republic. The essay includes a few references to structure and specific content from the book, so if something doesn’t make sense, it’s probably because it refers to an “insider” joke.

What’s inspiration is that my class comprised of people of all ages (including one who is 80+). Everyone was there for their hunger for learning, and I hope it stays with all of us until the very end of our lives.
PS: I’m the guy in the front and in the middle, the one who is wearing a white shirt.

I was sitting at my lunch table, eating bread and ruminating whether it was worth it for me to fly thousands of kilometers from India to Oxford, just to study one book: Plato’s Republic. ...  Read the entire post →

Don’t compete

The Internet is full of people winning all the time. Someone is traveling to exotic locations, someone else is raising funds, and another person is winning awards. Essentially, everyone around you is succeeding while you do spend your days as the nature intended – sleeping, eating, smiling, chatting with friends, and spending time with your cat.

But, who here is really winning?

Imagine our society as a living and breathing organism with its own agenda. What would be on its agenda? First of all, like you, society would want to not die. Second, like you, it would want to thrive. ...  Read the entire post →

2024 wrapped

This year’s review is going to be shorter than 2023 (and previous years) because I’m in Goa right now for a holiday and I don’t feel like being in front of a screen for long.

I mean, just look at this view and tell me that you’d rather be in front of a screen writing a review.

Photo by Aakanksha Gaur

But traditions must be upheld, so here’s a quick review of my eventful year.

🫡 Shut down Nintee

Earlier in the year, I shut down my startup Nintee and returned (the remaining) ~75% of funding back to investors. Everyone in my team got 6 months salary as a severance and an open offer to join my other company Wingify. ...  Read the entire post →

Why time seems to pass faster as we age

1/ I’ve been mega-obsessed with this feeling.

A year as a 36-year-old seems so much shorter as compared to when I was a kid or even as a teen.

It seems cosmically unfair – we have fewer years to live, and each year flies by faster.

2/ But, why is that happening?

My tentative conclusion is that it’s an unfortunate outcome of how evolution shaped our brain to be an efficient storage device. 

3/ Our brain is a prediction device.

Its top job is to construct a model of the world so that we get a survival and reproductive edge.  ...  Read the entire post →

Don’t sell your soul to the algorithm

The danger of pleasing the algorithm to go viral is that gradually you end up selling yourself to big tech companies.

This is how it works:

  • Algorithms optimize for time spent on platform, because more time spent = more time for showing ads
  • Algorithms promote content that sucks in more people (i.e. content that can go viral)
  • Certain types of content is inherently more viral (rage inducing, hot takes, lowest common denominator, etc.)
  • Creators maximizing reach prioritize creating such kind of content
  • Since creation shapes thinking (as much as the other way around), gradually they become what they tweet

    This loop has two sinister effects:

    • For the non-creator, it appears that the world is falling apart as they see extreme, hot takes all around them as nuanced, well-balanced content is seldom promoted by the algorithm
    • Creators with promise end up losing their soul in the process

    All this to make the richest companies and their shareholders even richer.

    This is why we must refuse to be dictated by the algorithm.

    It’s hard, very hard.

    But what is more important than likes and retweets is having an authentic voice (irrespective of whether it’s reaching to the masses or not).

    PS: It’s also worth noting that the algorithm forces dull sameness of content because machines optimize for a singular metric: time-spent. While what we need to become better thinkers (and also to save the democratic process) is diversity. ...  Read the entire post →

Bye 2022. Hello 2023.

This year was intense. Perhaps the most intense one in quite a while.

I’ve been gradually developing the habit of reflecting as months and years pass by. In my 20s, I used to think that celebrating birthdays or New Year is pointless. After all, what’s so special about Earth completing one revolution around its star?

Now in my 30s, I know that actually years are all we got. As I see my parents aging and grandparents not being around anymore, the relentless march of time is quite noticeable. I now fully understand that it’ll all be over and even though I can’t lock time in a bottle, I can at least bow and acknowledge as it departs. ...  Read the entire post →