In every life update post I keep saying that the past X months have been intense, and I wanted to start with the same, but I won’t, just to keep it different - even though the past two months really have been intense.
Guess I did it again, lol.
Well, I was a little tired of the fast-paced Bangalore life and had moved back to my hometown for a few weeks. As I write this, I’m heading back to Bengaluru. A lot of things happened, and it’s time to reflect.
Main track: Stripe & Legacy Systems
I’ll finish 8 months at Stripe soon, and I like to think I understand the Risk systems better now. Obviously there’s a long way to go, but I think I know the ecosystem well enough to have an opinion about things, and a decently informed one. The past two months, especially the last month, it has been about the good old struggle of ramping up on legacy systems that I had absolutely no clue about. I was leading a completely horizontal project, on code and processes, owned by over 8 teams - things that neither me nor my team had worked on before.
Working on legacy systems, as engineering connoisseurs, is challenging every time, since you encounter unique patterns and years of tech debt - which takes a while to absorb. The core mindset difference between operating in a legacy setup versus a relatively new, fast-evolving system is the amount of empathy and patience you have to bring to the table. It’s easy to get frustrated about why on earth would people build things this way, and it’s easy to take a moral high ground of “how I’d have done things differently.” But we often ignore how many iterations, urgent bugs, and incidents this system has been through.
One funny pattern I’ve noticed is that whenever a new person joins a team, their first obsession is “quality”. When I was five years into my first team, I was always mystified by this. Whenever a new person would join, especially a senior one, they’d start yapping about quality and start preaching best practices that everybody knows about. Funnily, when I joined a new company - twice - I subconsciously did the same. And I must admit that it’s just the inertia and discomfort of working in pre-existing patterns and tech debt, especially tech debt that we didn’t take on ourselves.
Most people are fine with tech debt, as long as they’re the ones who took it. If anyone else took it, suddenly there are quality wars.
Tech debt is a part of life. Just like gravity isn’t exactly 9.8 m/s² all the time, and just like you can’t be happy all the time despite doing the right things. Most of us start with great architecture that we envision scaling for years to come. And then life happens. The universe happens. An incident comes up where ten thousand users get blocked and you have to unblock them with a patchy fix. A $5M enterprise client shows up and you get them running in two months with a shitty architecture that “just works” but violates your initial semantics. Sure, as principled software engineers we can push back, but we always end up at some middle ground that involves some amount of tech debt.
So, my friend — there’s a high chance this “shit” legacy system you’ve been crying about would have been equally “shit” had you been the perfectionist initial architect and tech lead for it.
That said, I don’t mean to say we stop trying and pushing the quality bar. We absolutely must. That’s all software is about: keep debating with others about how their tech debt is degrading the semantics, while our tech debt is helping us increase revenue.
Fat loss
At 89kg now, 3kg down in 3 months. Extremely slow, because moving back home makes it annoyingly hard to resist the delicious mom-made food you don’t have access to when you live away. The mutton rassa, fish curry, yellow varan bhaat, vaanga-bataata, homemade pickles. None of it is remotely healthy, but it’s hard to say no every time.
Don’t have my health numbers because I lost my Ultrahuman ring last month and got a new Garmin just two days back. I’ll track and post the numbers in the next update. Garmin feels like a slick ecosystem — decently consistent, accurate, and good battery life. Ultrahuman was absolutely dogshit. It showed my heart rate at 100bpm after a squat PR, and one day it clocked me at 200 when I touched my pulse and it felt like barely 120. Unreliable doesn’t even cover it.
All in all, as I mentioned earlier 0, fat loss is frustrating as always and I have 5 more kilos to go, but I’m happy I didn’t gain much at home.
Passion Tracks
Stopped building the AI browser companion because I haven’t had the time, Claude Code has been quite good for all things AI assistance, and honestly I’ve been a little lazy about it. But there’s something deeper too. As much as I think AI is useful and revolutionary, I’m a little irritated by the unnecessarily fast pace of everything right now. The sense of urgency at work is good and I genuinely enjoy it, but the Bangalore life had quietly become something else entirely — work, social events, table tennis prep, gym, weekend treks, repeat. One evening I came back after work, lay down on the sofa, and just stared at the ceiling. Then friends showed up and we had to go out. I realised I hadn’t had a single moment to just empty my mind and get properly bored in weeks. Three days later, I’d booked a flight and left for hometown. No plan, just out.
As I’ve mentioned earlier, I’ve been doing some engineering collaboration with Step Up For India 1 (SUFI). We’ve been working together for almost 10 months now. Shipped some website work, and now we’re working on intelligently creating videos from English textbooks to help children learn English. If this works out well, I’ll talk about it at length in a separate post. Overall, pretty excited about getting this through the finish line — it will be extremely impactful.
Also started working on Enfield, my first nuclear engineering project. The idea is to index nuclear data and build a natural language interface on top of it — parse and index ENDF, the Evaluated Nuclear Data File, a public but deeply arcane library containing cross-sections, decay chains, and fission yields for hundreds of isotopes. On top of that, a computation layer that can derive answers at query time, unlike regular RAGs. The goal is to ask Enfield something like “what happens to xenon-135 concentration six hours after a PWR scrams” and get a short, cited answer backed by real nuclear physics. Think Wolfram Alpha for nuclear data, with clean software architecture underneath.
Nuclear physics was the most boring subject I encountered in college, but I’ve since realised that’s almost never the subject’s fault — it’s the exam-oriented way we were taught everything. History was the same: memorised question-answers, zero context, zero story. Turned out I love history. Spend most of my free time reading and watching about it now. So nuclear physics felt like unfinished business — a subject I never actually met, just survived. That, and I genuinely want to see what serious software engineering looks like when applied to a domain that actually matters at civilisational scale. Nuclear energy is going to be significant in the next few years, given where energy and compute demand is headed.
I know barely enough nuclear physics to be dangerous. That’s the point.
What next
Besides the exciting stuff at Stripe that I obviously can’t talk about, I’m looking forward to three things.
Going even more aggressive on fat loss — 5 kilos in the next two months. It’s going to be hard, especially with a two-week vacation coming up in mid-April.
Speaking of the vacation, that’s the next thing I’m really looking forward to. Past three vacations have been treks, but this time I’ll just be going to Uttarakhand and vagabonding my way around - it’s been a while since I’ve lived the backpacker life. Just a backpack, no agenda. Start in Rishikesh, maybe rent a motorcycle and ride to Gangotri, or just wander through nearby villages. Let’s see.
And finally, by the end of May, I want to finish building Enfield and make a real entry into nuclear physics.
Life lately has felt like a river running a little too fast - beautiful, but hard to see clearly. Slowing down, even briefly, helps you remember what you’re actually moving towards.
References
- [0] - Fat loss: What to expect - https://rishi.monster/posts/fat-loss-what-to-expect
- [1] - Step Up For India - https://stepupforindia.org/