Shreyansh 🐾🕊️

आराम कमाने निकला हु , अपना आराम छोड़कर ।

Cancer is cured by AI. GitLab founder Sid Sijbrandij was diagnosed with stage 4 spinal cancer. Every trial rejected him. His doctors had nothing left to offer. So he stopped being a patient. He built an AI research team. Fed them 25TB of his own medical data genomics, scans, treatment history, everything. The system found a treatment his entire oncology team had missed. Then engineered 19 custom vaccines from his own DNA. Relapse-free since 2025. Then he uploaded the entire blueprint. Free. For every person sitting in that same room, hearing the same verdict, with nobody left to call. Medicine runs on averages. AI runs on you.

Two-phase commit (2PC) gets talked about a lot in system design, but how about coding it and seeing it in action? Here's me explaining 2PC through and through... I walk through not just the concept but a full code of 2PC on a regular SQL database -- the coordinator logic, the prepare phase, the commit/rollback, all of it. No fluff, no black boxes. The truth is, 2PC is not that hard to implement. What is hard is reasoning about what happens when a node crashes between the prepare and commit phases, or when the coordinator itself goes down. That is where the real complexity lives. This is precisely what I talk about in this 40-minute video. This is what system design actually is -- not drawing boxes with arrows, but thinking through the code and the failure modes at every step. Give it a watch.

Interviewer: An OTP is valid for exactly 30-60 seconds. Even though no one stored it anywhere, how does the server verify it without ever saving it?

Hugh Jackman stars in the official trailer for #TheSheepDetectives. The story of a murdered shepherd and his mystery-obsessed flock who must solve the case.

They built an AI so powerful, that they had to k*ll it.

AI Engineering is exploding. 🤯 But here’s what most people don’t realize: The real bottleneck in AI today isn’t training models. It’s running them at scale. That field is called Inference Engineering. And it’s becoming one of the highest-demand roles in AI. Here’s how to break into it 🧵👇

There’s a lot of fear-mongering right now about software engineers being replaced by AI founders and AI companies. A massive amount of money is being invested in this space, and naturally, companies need a strong narrative to justify high valuations and keep the stock market excited. Today a company laid off 40% of its workforce citing “AI efficiency,” and its stock jumped 24% in a single day. Later, after public backlash, the founder clarified that he had significantly overhired during the COVID boom and were simply correcting their workforce size. Many of the mass layoffs we’re seeing today are not purely because of AI replacing engineers. Most of them is due to overhiring during COVID era. Current AI systems are not capable of fully replacing software engineers. Yes, they can generate impressive code. But engineers are not hired just to write code. Will AI eventually eliminate all software engineering jobs? I don’t know and I’m not qualified to make such a definitive prediction. What seems more realistic is that AI may reduce the total number of engineers needed, because individual developers are becoming more productive with tools like Cursor, Claude Code, Codex, etc. Productivity gains don’t necessarily mean zero engineers; they may mean leaner teams. No serious company today is going to give an AI full autonomy to make critical decisions and deploy directly to production without human oversight. The risk of technical, legal, financial, and reputational are too high. It’s better not to get swept up in the fear amplified by X echo chamber. AI is a powerful tool, and it will reshape software engineering but reshaping is not the same as wiping it out.

> be Tanay Kothari > spawn in Delhi > attend DPS RK Puram > yes, that DPS RK Puram > the legendary school that casually pumps out $100M+ founders (boAt, Snapdeal, Sugar, Cartesia) > become a literal academic anomaly > score a flawless 36/36 on the ACT > complete 17 undergrad CS and Math courses from Stanford, MIT, and UPenn > ...before even finishing high school > casually master 20 programming languages by age 16 > grow up faster than standard education can handle > grind out god-tier side quests > become the youngest Microsoft Certified Solutions Developer on the planet > win medals at the Asia Pacific and International Informatics Olympiads > write a paper diagnosing early-stage cancer via image scans > lead a 5-country engineering team for a NASA space settlement project > all while other kids are playing video games > head to Stanford for CS & AI > don't just sit in lectures > end up teaching the university's Deep Learning course alongside AI legend Andrew Ng > publish cutting-edge research at the Stanford AI Lab > build and sell an AI startup, FeatherX, before even taking graduation photos > look at the state of voice AI and realize it's an absolute joke > giga-pivot to killing the keyboard > build Wispr Flow > idea: let people literally talk to their computers at the speed of thought > achieve sub-700ms latency and an 85% zero-edit rate > models so good that Siri, ChatGPT and Claude feel ancient in comparison > build native support for developer workflows so you can literally code by talking > hustle like a maniac > personally onboard the first 500 users via Google Meet > Reid Hoffman, Steve Wozniak, and Marc Andreessen get instantly addicted > every VC in Silicon Valley is using it daily before the Series A pitch even starts > secure $81M from Menlo Ventures, Notable Capital, and 8VC > hit a massive $700M valuation > rewriting how humans interact with machines

Coding nowadays

Deleting a large number of rows (say, a million) in a single query feels efficient. It is not. It can crush your database. When we run a large DELETE in one go, the database has to hold locks on every row it touches for the entire duration of the operation. If it takes 30 seconds to run, those locks are held for 30 seconds. Every other query that needs those rows is stalled, waiting. On busy systems (high query or update load), this has a ripple effect. Reads pile up, writes get blocked, and your connection pool starts getting exhausted. This cleanup job will likely become a production incident :) There's also the transaction log (WAL) to think about. A massive DELETE generates a huge amount of log data in one shot, which can spike disk I/O and slow down replication. Your replicas can fall behind, sometimes significantly. The fix is pretty simple - batch your deletes. DELETE ... WHERE ... LIMIT 1000, then sleep for a small interval, then repeat. It's slower in wall-clock time, but the locks are short-lived, the log writes are spread out, and your database stays responsive throughout. Fun fact: Databases cannot protect you from yourself :) This is one of those things you learn once, usually the hard way. Don't ask me how I did :)

You will get what you dream of; just consistency and hard work are needed.

I am creatively ovulating

🕔There are 24 hrs in a day, and 7 days a week. Find a consistent schedule that works for you. Extract 15-20 hours in a week for yourself. E.g 5-7am X 5 weekdays + 1 weekend = 20 hrs already. ˜1000 hrs for yourself in a year. Enough can be done.

10-minute delivery in India isn't magic. It's really good engineering. @albinder and @letsblinkit built tech that most people never see. We went deep on how it works.

I come back to this speech every week

Next decade secured and we’ll be there! 😀 🍿

🚆 Train No. 09057 is running very late today, causing inconvenience to passengers. Requesting @RailMinIndia @IRCTCofficial to please look into this and provide an update on the delay. #TrainDelay #IndianRailways