CEO @fastcodeai | Research-grade AI for enterprises | Dad

Bangalore, Karnataka, India
Joined March 2012
I'm looking for a younger me. The current me has a PhD from Max Planck. Taught at IIT Bombay. Built tech at Apple and Weta Digital. Published papers with Yann LeCun. But my most important lessons? They happen between meetings. In Uber rides. Over coffee with founders. And they vanish into thin air. I forget. Balls get dropped. That changes now. I'm hiring my shadow. My right hand. Someone who'll be in every room, every call, every decision. You'll manage my calendar and keep me sane. It's a Chief of Staff role. Think of it as a real-world MBA. You'll see how we close deals, handle crises, build partnerships. You'll learn why some meetings are worth ₹2Cr and others aren't worth 2 minutes. My last Founder's Office hire Rithika Gopikrishna is now a Chevening scholar. ₹6-12 LPA plus an education you can't buy. In-office 5 days, Bengaluru. I don't care about your degree. Hell, I don't care if you finished college. I care if you're hungry. That's the younger me I'm looking for.
7
1
1
25
1,957
I asked AI to set up a LinkedIn ad for Restart using computer use, our AI course for people restarting their careers. It took 86 browser actions. It tested audiences. It built 2 versions of the ad and priced both. It found a LinkedIn tag already on our site, so our developer had almost nothing to do. It wrote the lead form. Then it asked me 5 questions. My replies: "you decide" "yes it is" a link "no, it's fine" "13th november" 86 actions for it. 5 lines for me. This is what AI is for. The boring, tedious work. Not the thinking. Not the creativity. Who the course is for. What it promises. Whether it should exist at all. That was ours before AI opened a single tab. Nothing goes live until I say go. Keep the thinking. Hand it the clicks.
4
9
362
So cool! AI is sniffing out age old bugs that we were living with!
Hi all, I have a new result I'm excited to share. It turns out that *all existing implementations* of softmax and cross-entropy (that's JAX, PyTorch, textbooks, etc) have a subtle bug --- two bugs, actually. A simple rewrite fixes both, and eliminates "slingshots" when grokking:
1
4
613
Arjun Jain | Fast Code AI retweeted
Ben Horowitz's message to early-stage founders on betting on yourself, the doubts that do the most damage, and why fear means you're doing it right: "Your job is to make your bet pay off. You believed in yourself enough to do it. There are gonna be many times when you question that bet and you question that belief, but that gets you nothing." "The doubts are gonna be there. Ignore the doubts. The worst doubts aren't the ones people put on you. It's the ones that you put on yourself. Every entrepreneur has them." "I spent a lot of time with Mark Zuckerberg when he was 22 years old and he had a tremendous doubt about whether he was gonna make it." "It's not what you feel. It's what you do. You're gonna feel scared. If your guts aren't boiling, you're not even trying at this thing." @bhorowitz w/ @speedrun GP @andrewchen
66
128
57
1,152
140,492
Move 37. AlphaGo's intuition put the odds a top human would play it at 1 in 10,000. AlphaGo played it anyway. Its search looked further ahead than its intuition. Thore Graepel helped build AlphaGo. This month he wrote that today's LLMs has the intuition but not the search. And that its chain of thought is often a story told after the fact. We measured t  Some driving AIs now explain themselves. "I slowed down because a cyclist was merging." In CausalDriveBench, our NeurIPS 2026 paper, we tested whether the explanation is why the car moves. In one test, we removed the reasoning. The trajectory didn't change. We swapped in reasoning that contradicted it. The trajectory didn't change. 13 models. 7,285 questions about cause and effect. How well a model answered the why-questions said nothing about how well it drove. 4 of the 13 scored below random chance. The reasoning is ornamental. In a car, that matters. When something goes wrong, the explanation is the first thing an engineer reads. If it isn't why the car moved, it's just a story. Graepel is right. Reasoning has to be load-bearing. In the models we tested, it isn't. Yet. Vision and sponsorsip by Aish Dubey and joint work with his team at Renesas and also IIT Delhi.
1
8
498
"Whatever we think life is, that's what life becomes." Tony Robbins, yesterday. We don't see the world as is. We see it through a lens. We label before we look. Good or bad. Friend or threat. Win or loss. My lens: everything is a fight. A minor bug looks like the sky is falling. A family board game looks like a qualifying lap. Same bug. Same game. The lens did that. Jains have a name for the lens: mohanīya karma. Perception-deluding karma. It puts a color on everything you see. Tony says change the color, and life changes. Jains go one step further: take the lens off. I'm still working on a better color.
Whatever we think life is, that's what life becomes.
1
7
420
Can’t wait for the model to drop. Would love to also see a technical writeup or training code (asking for too much now?)!
We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence. block.xyz/dessa
3
473
Starting a company to become rich or famous is indeed super silly. Being an entrepreneur forces you to be the best version of yourself. For example, if I weren't running a business, I would never write on social media.
1
7
495
AI will distinguish quality from jugad. I am happy.
Gross incompetency from Indian certificate authority eMudhra. You had one job: generate a 384 bit private keys and instead you used: 4, 56, 66, 115, 155, 156, 230, and 237. This allows attackers to issue certificates for any websites. Embarrassing that it’s a family business run by largely computer illiterate people. Only 1 of the 3 have a remotely computer-related background and even they never actually worked a technical job in their life. If you look at the LinkedIn, their tech team is young, from tier 3 education and paid sort of bottom barrel. Imagine how many attacks you can design with AI given the existence of this genre of gross negligence.
1
2
20
920
Tesla is dangerous because it oversells. What the fans believe and want us to perceive doesn't change reality. nypost.com/2026/10/16/us-new…
2
8
736
Last night I watched Schumacher 94. As a boy, he only had access to a karting track because his mum worked there in the canteen. He raced on tyres he pulled out of richer kids trash bins and learned to win on bald rubber. And because of this, in the rain, nobody could ever touch him. James Allen sums him up cleanly: They have this rage to compete... Obviously, its extremely difficult, its extremely dangerous, but thats part of the attraction. The more difficult something is, the greater the satisfaction that they get from it. Nobody is born with that rage. I was 7 when my parents left me at a boarding school in Indore. 16 boys to a dormitory. You learn fast that nobody is coming to save you. That chip on my shoulder got me through a PhD in Germany when I couldnt speak a word of the language. It got me in front of Mercedes with a demo. It built a company of 50 from nothing. In your 20s and 30s, that rage to compete is a superpower. In your 40s - if you havent done enough therapy - that exact same fire makes you exhausting to live and work with. Because when you are conditioned to only feel alive in extreme difficulty, you treat everything like a street fight. At work, you hold people to unrelenting standards because good enough feels like a threat to survival. At home, a quiet weekend feels suspicious - if there isnt a problem to solve, your brain quietly invents one just to feel useful. You end up treating a minor bug like the sky is falling, and a family board game with your 8-year-olds like a qualifying lap. Even when you’ve built the company and earned your seat at the table, your nervous system doesnt get the memo. You’re still driving like you’re on bald tyres in the rain. I learned how to win the hard way. I still have no idea how to do the easy way.
2
1
20
1,011
Let me be blunt: the German Mittelstand companies that embrace AI will keep their competitive edge and move forward. The ones that don't will fizzle out and die, no matter how good their engineering is. Don't get me wrong, the Mittelstand has serious expertise, and these companies are excellent at what they do. Decision-making is also quick, because you're speaking directly with the owners, in this case a father and son. But succession planning, staying competitive and holding off global players are real pain points, and that's where we're now helping them with AI.
1
11
586
Cute demo, no doubt. But like @ylecun we always ask: Will it work in the real world or is it designed just to create hype?
Replying to @ylecun @elonmusk
Okay, so it's just compressed air. Easier. But crappy specific impulse (less than 50 seconds). Even assuming that 50% of the mass of the car is the compressed air, that gives you less than 10s of useful thrust (with thrust-to-weight ratio larger than 1.0). Probably more like 5 seconds, actually. So the car must be an empty, lightweight shell containing big compressed air tanks. Cute demo, but totally impractical. But of course, you knew that.
5
15
867
I think people are underestimating just how much industrial knowledge Germany is sitting on, and how easily it could lose its competitive advantage over the next decade. Germany has spent generations building some of the most sophisticated industrial companies in the world. Many are family-owned Mittelstand businesses, operating in obscure niches, making products that most people have never heard of, but which entire global supply chains depend on. Their competitive advantage was built over decades. It lives in the accumulated experience of their engineers, the intricacies of their manufacturing processes, and an intimate understanding of problems that competitors may not even know exist. And yet, these companies face an increasingly uncomfortable future. Energy is expensive, regulation is burdensome, skilled workers are difficult to find, and competition from China is intensifying. Meanwhile, technological change threatens to erode advantages that took generations to build. I find it troubling how casually we discuss Germany's industrial decline, as though we are talking about abstract economic statistics rather than businesses that families have spent their lives building. But I also believe AI offers these companies an extraordinary opportunity. Consider the sheer quantity of knowledge accumulated inside a fifty-year-old manufacturing business. Engineering drawings, maintenance records, production data, and decades of experience troubleshooting problems on the factory floor. Much of this knowledge remains fragmented, difficult to access, and dependent on a handful of experienced engineers. Imagine making that accumulated expertise accessible across the organization, allowing engineers to understand complex dependencies and make better decisions in a fraction of the time. This is precisely the kind of problem we're working on at @FastCodeAI. Germany's greatest industrial strength is its accumulated engineering knowledge. Air-gapped secure customized AI offers an opportunity to make that knowledge considerably more valuable. And I believe preserving this advantage should be a national priority.
3
3
3
37
3,533
Young folks who want to succeed, work life balance is a myth. Life Jeff Bezos calls it, it can be work life harmony. Example, check the requirements from Khosla’s job post: •Work in person in Menlo Park every day, generally from 8 a.m. to 8 p.m. •Maintain around-the-clock availability, including nights and weekends when needed. •Commit to three years. •Live near Menlo Park or relocate there. Relocation support and visa sponsorship are available.
I'm hiring my next Chief of Staff. You won’t know every field I work in. What matters is how quickly you learn, the questions you ask, and whether you can think from first principles. You’ll help evaluate opportunities and work with founders building consequential companies. Technical background preferred. Three-year commitment. Menlo Park, in person. Apply: jobs.ashbyhq.com/khoslaventu…
10
10
174
51,925