Posts

You're on the same team

Image
To human is to mis-express and misunderstand.  Here, poor Pam went from feeling romantic, to Jim expressing underlying desires. They both want to feel love, just in different ways. And if they miss recognizing they're both on the same team, they're in for some solid fiiights. Why does this happen?  I think it's because: Most of us don't know what we feel underneath.  So our underlying feelings come out through trivial everyday conversations. We snowball very quickly, and we don't know how to catch it quick. We're never taught to step back and think why we're fighting. We jump to conclusions quickly. And assume the other's intent. And almost always, I'd say we're wrong. There are 8 billion worlds on this planet. When two people interact with each other, it is like two worlds interacting. Water splashes from one, causing gentle rains on the lands of the other. Some storms, some sunshine. Sometimes, tectonic plates collide. We have two ways to inter...

Why relationships are hard

Image
Jim and Pam are about to leave the house to get brunch Jim feels silly because this could've totally been prevented. " Car keys , ah, why didn't I say car keys", he thinks to himself. Pam feels silly because she could've asked "yep, but you got the car keys right".  Because obviously since she was leaving the house last, he meant the house keys. A couple days later, after their big move to the new city, Jim and Pam are chatting.. Now this has the potential of becoming resentment if Jim and Pam don't clarify what they really mean immediately. "You always complain even when I put in effort" says Jim a couple weeks later. "It's because you never use the right words" exclaims Pam. "I just wish you understood what I'm saying. I never mean to attack you" replies Jim. That. Look at how it snowballed from just a little bit of dust. But there's an even more peculiar, more common form of misunderstanding. One where you...

You can't control the weather

Image
There's a mountain peak you've been wanting to go. But going there requires leaving your house, leaving your safe space. You'll be alright if you don't go. The mountain peak was just curiosity.  The house is safe, you know exactly what you're going to get. The safe shelter of your four walls, the warmth of your blankets. But. The mountain peak. There's this voice in you, wanting to come out and experience the unknown. You say to you yourself, "I was built - even if just metaphorically - to explore. And not doing so troubles me". You can shut that voice, until there is no desire to do anything at all. It's wild and scary out there, and you don't know what you're going to get. Yes.  That's it. The safety of the indoors over the unknown of the outdoors. So, is that it? Either the comfort of the house, or the vast unknowns of the outside? No. Recognize there's a voice. Feel the longing for whatever it is from within. Learn about that jo...

Living in alignment

Image
I've learnt that one of the biggest sources of stress is living in misalignment to the way you would rather be living. When I was at Google, I wanted to increase my ownership and say at the table. After years of trying, I realized some organizational things just can't be changed. I lived in Germany, but the decision makers were in America. So even though I never overworked, always had brilliant colleagues, ate wonderful meals at Google 3 times a day, there was an overarching theme of stress. Of me being in a place I wasn't able to fulfill my desires. Then, I joined Revolut. I got to own really large, challenging problems. I got to work on problems on my own terms and solve them how I'd like to. And even though I overwork at Revolut, I rarely feel stressed. I'm happy because it is in alignment to how I want to be working. I guess in other words, do shit that makes you happy :D

How I view LLMs as Sept 2026

I’ve spent over $30k in tokens since the start of the year across Codex, Claude, and GLM. Here’s how I view LLMs as of September 2026. Two categories: Instructions-following models: Like Luna and Sonnet. I used to call these “dumb”. But that’s the wrong way to think about them. They just cannot make good “common sense” decisions. They’re workhorses for pennies. Luna especially, is hard to spend a lot on. If you make the core decisions up front, coding is auto-complete. Intent-understanding models: Frontiers like Astra and Fable. These are higher “common sense” models. These models (try to) understand the intent behind the question. Solving problems is about making the right tradeoff - these models tend to make the right tradeoff more often than the other models. These models know how to orchestrate instruction-following models prescriptively - in a way humans get too lazy to. In my experience, today’s models are far from making the right human-like trade-offs. However, this is where I ...

Uninterrupted Hours

Image
One doesn't stumble into greatness. Everything great takes solid effort. And effort takes hours. But where do you find those hours? The human schedule is full of things that keep us away from putting in those hours. Eat, exercise, hygiene, social, mental well-being, you name it. A usual day (at best) in most of my software engineer friends' lives. When I was in 11th and 12th grade, my parents made it possible for me to just study, and care for nothing else in the world. Those were the years I put in the most hours to a discipline than I ever have. There was no concept of study life balance . That study life balance was to come after 2 years of this solid grind once you got admitted to a top tier university. Until then, weekdays meant school all day and study all evening. Weekends meant study all day and study all night. And I miss that. - The reason most of us never get to put in any solid effort, raw hours, in pursuit of getting really good at something is because humans an...

Think in user journeys

You're tasked to design a hotel room. You start with a bare room, and you have to turn it into a fully functioning hotel room. What do you do? From memory, you know what hotel rooms have: a bed, a toilet, a cupboard, a shower. All the bare minimal functional things. You remember you like the bathmat at a hotel you stayed at a year ago.  You put it all together and you get a hotel room. It works, but it's really just the basics.  It solves the the obvious problems, but it misses the actual experience. When thinking from memory, or without structure, you miss out on small details that make the experience go from being the bare minimum to very pleasant.  Specifically: A guest opens the door - what is the first thing they do? They probably look for the light switch - can they find it without thinking? They dig in their luggage to find a change of clothes - where do they place their baggage? What is their shower at this hotel like?  They need a place to keep their toiletr...

8 Billion Right - My mental model of the world

Image
There’s a Indian tale of an elephant and 6 blindfolded people. Each is made to feel a different part of the elephant. One touches the bushy little tail and shouts, it’s a rope! Another touches its massive leg and shouts it’s a tree! One says it’s a spear, another says it’s a wall, and so on. Each person comes up with a different answer, based on the perspective they’re made to look at the giant from. Now you could say they’re all wrong, but that’s not true. The lesson I’m drawing: given the information they have, none of them is wrong. Each, in their own reality, is right. Life is much the same. If “life” is a massive blob filled with colors and objects and emotions that everyone gets a different pinhole to view and experience from, we’re all going to have different perspectives. Each person will establish their own ideals, belief systems, and rights-vs-wrongs. So everything that anyone ever does, is right . (right != morally correct) It is the right thing to do given their world, give...

The last line of code I ever wrote

Image
June 2025 was the last time I wrote code by hand. And it is the last time I will have ever written code by hand in my life . Yikes! That's taken me a few days to sink in.  I remember my undergrad days when I'd code all night building Webmail  and Lifehacks .  I'd read up every design pattern, every article, scour StackOverflow for every question that remotely related to what I was doing, and naively try to apply it for my use case. In retrospect, I was learning how to build long term maintainable projects.  It was difficult, most examples I found online were either too easy, or too difficult to follow. It was an arduous, gradual process, but I believe that's what it takes for humans to get good at something. It's March 2026 now. 9 months since I last wrote code by hand. I have submitted over 5,000 lines of code at the company I joined a few months ago. I have made over 2,000 commits to personal projects . And every single one was written by an AI. I'm not vibe ...

Special is what you make it

Image
There's this scene in Kung Fu Panda where Po asks his father - the goose who runs a noodle restaurant - what the secret ingredient in his renowned  Secret Ingredient Soup is . I expected the usual answer: love. But he says probably the wisest words I've heard in a movie: to make something special, you just have to believe it is special. (I absolutely love this movie) And that's it. Special is what you make it. That's new years eve. The sun rises the same it always does. Nature doesn't mark itself on this specific day celebrating another trip around the year. Humans just collectively decided that this is the day. We celebrate, we light firecrackers, we wish each other happy times.  What you believe to be special, is special. That's what traditions are. That's your birthday. That’s the exam you studied for two years for. That's the first day of your new job. That's the last day of your previous job. That's the day you get married. That's the da...