No teacher in sight.
That turned out to be the point.

Sandeep Giri grew up in a tiny, remote village in India. There were no exceptional teachers nearby, no coaching centres, no curated curriculum. What there was — was books. And books turned out to be enough.

Most people would describe this as a disadvantage. Sandeep doesn't. Not having a teacher puts you into first-principles learner mode by necessity. When no one is there to hand you the answer, you have no choice but to sit with the problem until you find it yourself. You cannot wait to be taught. You cannot outsource the thinking. You learn — really learn — because there is no other way forward.

And because you have to do everything yourself, you stay hands-on — not as a method you choose, but as a habit you cannot shake. Theory without practice never becomes comfortable because it was never available. The only knowledge that felt real was knowledge you had put to use.

There is a third thing that grows quietly in this environment: you make a habit of blaming no one. When there is no teacher to blame for a poor explanation, no system to blame for a bad curriculum, no circumstance to point to — you stop looking outward for the cause of your own confusion. You look inward. You find the gap. You fill it. That posture, formed in a village with no coaching centre and no mentor, never left him.

He read voraciously, and when a book genuinely moved him, he went further than most students would think to go. He wrote to the authors. When H. C. Verma — the physicist behind Concepts of Physics, the book that defined how a generation of Indian students learned to think about science — received a letter from a village student with questions that went beyond the text, it said something about how Sandeep had always engaged with ideas: not as content to absorb, but as conversations to continue.

When Sandeep reached college, something clicked. Computing wasn't just another subject — it was a medium for invention. While classmates attended lectures and moved on, he studied every aspect of computing: operating systems, algorithms, data structures, computer architecture, networks. Not for grades. Because each idea led to a deeper one.

He dug particularly deep into computer security — the layer where theory meets real-world consequence. He became one of the early contributors to research on XSS (Cross-Site Scripting) vulnerabilities, publishing findings on BugTraq, the security mailing list that serious researchers read in the early internet era. This wasn't a side project. It was the same instinct he had always had: go further than the course asks.

He also co-founded IMG — the Information Management Group — which built his college's website, online registration system, and digital infrastructure. At a time when most students were still figuring out how the internet worked, Sandeep and a small team were building it for their institution. It was his first experience of shipping something real that thousands of people depended on.

In 2002, Sandeep joined D.E. Shaw, one of the most selective quantitative investment firms in the world — a company that hired fewer engineers than most IIT departments graduated. Getting in required exactly the kind of thinking he had spent college developing: the ability to reason from scratch, not recall from memory.

What he worked on there wouldn't acquire its now-famous names for another decade. He was doing deep learning and big data — building systems that learned from vast datasets, that found patterns no human analyst could see — years before these became the defining technologies of an era. The vocabulary didn't exist yet. The problems did. And Sandeep was already solving them.

But something else was already alive in him: the compulsion to teach. Even at D.E. Shaw — surrounded by some of the sharpest technical minds in the world — Sandeep wanted to run sessions on how to write secure code. The catch: his audience were top scientists who had no reason to sit through a lecture from a younger colleague unless he could earn the right to teach them.

So he did what he had always done — he went to first principles. Before standing up to teach, he demonstrated actual bugs in their existing systems. Live. Undeniable. The room paid attention. The credibility wasn't granted by title or tenure; it was earned by showing, not telling. It was the earliest version of a principle that would later define everything he built: demonstration before explanation.

The entrepreneur in him couldn't stay dormant. Sandeep founded tBits and set about solving a problem that was far less glamorous than fintech or social media — and far more consequential. He built a document management system that became the backbone of engineering and compliance workflows across India's power sector.

Most of the power plants built in India run on software Sandeep designed at tBits. Every engineering drawing, every regulatory document, every maintenance record — managed through a system he architected from first principles. It is the kind of infrastructure that disappears into the background precisely because it works so well. A reminder that the most important software is often the software you never see.

At InMobi — one of India's first global tech unicorns — Sandeep built the recommendation engine that decided which ads reached which users across hundreds of millions of mobile devices. The system processed 200 terabytes of data to make those decisions in real time.

At that scale, the difference between an elegant algorithm and a naive one isn't academic — it's the difference between a system that runs and one that collapses. He saw, firsthand, how the most impactful engineers weren't the ones who knew the most frameworks. They were the ones who understood the underlying mathematics well enough to know when the framework was wrong — and confident enough to replace it.

At Amazon, Sandeep owned the mobile version of Amazon's product detail page — the page from which hundreds of millions of customers make purchases every day. One of the highest-traffic, highest-stakes surfaces in e-commerce.

Among his contributions: he redesigned the image selection framework — the system that decides which product image to show to which user in which context — achieving a 5× performance improvement through first-principles thinking. Not by tuning parameters or swapping libraries. By going back to the underlying problem and redesigning the approach from scratch. Exactly the way he had always worked.

It was also at Amazon that a pattern he'd observed for years became impossible to ignore: the engineers who struggled weren't lacking intelligence. They were the ones whose education had trained them to reproduce answers — not generate them. Amazon's engineering culture is famously first-principles. The people who thrived there had, somehow, learned to think the same way.

Across D.E. Shaw, InMobi, and Amazon, Sandeep hired engineers, mentored new graduates, and watched the same pattern repeat itself. Brilliant students. Strong marks. Completely unprepared. They had memorised textbooks and survived exams — but when a problem had no answer key, they froze.

He understood why. These engineers had always had teachers. Every concept had been explained before they were asked to use it. Every question had been answered before it became difficult. Their entire education had trained them to receive knowledge — not generate it. When no one was there to hand them the next step, they had no idea what to do.

Sandeep had grown up in the opposite condition. No teacher meant no answer until you found it yourself. That habit — of sitting with a problem until it yielded — was not a personality trait. It was something his circumstances had built in him. And it was buildable in anyone. That was the gap. That was what he would fix.

Teaching your own son changes you. There is no professional distance to retreat behind, no authority to invoke, no bell that ends the session when things get uncomfortable. If he doesn't understand something, it is on you — not on the child. That is not a philosophy. Sitting across from your son, watching him struggle with an idea you just explained, makes it an undeniable fact.

And the standard is higher than mere understanding. If the child is not finding the teaching interesting, the onus to make it interesting falls entirely on the teacher. A bored child doesn't pretend to follow along out of politeness. He simply stops. Every teacher can hold a room through authority or obligation — but holding a child's attention through genuine fascination demands something else entirely.

This is the stress test that most teachers never face. Sandeep faced it at home, and it rebuilt how he teaches. The habit of watching for the exact moment understanding lands — or doesn't — the instinct to throw out an explanation that isn't working and find a completely different angle, the belief that making something interesting is the teacher's responsibility and no one else's: all of it was sharpened in that private classroom. It is now built into every session he runs.

For over a decade, Sandeep has given up his weekends to teach — at least eight hours every Saturday and Sunday, without exception. Not as a commercial obligation. Not as a branding exercise. Because he genuinely cannot stop.

He believes something most people only say: everyone must teach. Not because it helps students — though it does — but because teaching is the one activity that forces you to find the gaps in your own understanding. You cannot hide from a confused student. The moment they don't follow, you discover exactly what you had never truly understood yourself. Teaching purifies your knowledge.

When a student is stuck, Sandeep doesn't repeat the same explanation louder. He reframes the problem — finds simpler words, or breaks it into smaller pieces that are each within reach. The goal is never to make the student feel the problem is hard. The goal is to make them feel the problem is solvable. Because it always is, if you find the right angle.

But the moment they solve it — even a small piece of it — everything changes. When a student is struggling, nudge them silently. A quiet hint. A reframe. A gentler version of the problem. No fuss. No drawing attention to the gap. But when they solve something — celebrate it. Loudly. Make it impossible for them to miss the fact that they just did something real. This is not performance. It is precision. A student who feels the weight of their own success starts to respect themselves as a learner — and that changes everything about how they approach the next hard problem.

The third principle follows from the first two. If you want to truly mentor someone, you have to understand them well enough to find their strengths — and then hold those strengths up where they can see them. Not false praise. Genuine recognition of what is already there. A student who knows their own strengths becomes a different kind of learner: curious instead of anxious, persistent instead of avoidant. That is the student who keeps going when things get hard.

Most teachers, when a student fails to understand, reach for the same explanation: the student wasn't paying attention, didn't do the preparation, isn't cut out for this. Sandeep has never believed that.

His view is unambiguous: if a student cannot understand something, the responsibility lies with the teacher — not the student. The explanation was wrong. The sequence was wrong. The example was wrong. Something in the teaching failed, and it is the teacher's job to find out what and fix it.

This isn't modesty. It is a design constraint. It forces a teacher to keep finding new ways to reach people — and it's the reason his students describe explanations they have never forgotten, on topics they had given up on understanding.

The first thing Sandeep and Abhinav agreed on: no passive video watching. The platform they built required learners to actually write code, run it in the cloud, and get feedback — immediately, without setup. No local environment. No installation friction. Just a problem, a browser, and the requirement to solve it.

This was 2014. Browser-based cloud labs for learning to code were not yet standard. CloudxLab was one of the first to make them work reliably at scale.

Learning to code is hard. Most people quit not because the material is too difficult, but because the feedback loop is too slow. You put in effort; you don't know if it's working; you lose momentum.

Sandeep and Abhinav added gamification components — streaks, badges, leaderboards, instant automated feedback — to make progress visible and the next step obvious. The goal was not to make it easier. The goal was to make it impossible to stop.

CloudxLab grew. The platform worked. Students were learning. But Sandeep kept noticing the same thing: students who completed the platform could use tools, but still struggled when a problem looked different from what they had practised. The platform was teaching skills. He wanted it to teach thinking.

Over years of iteration — watching students, adjusting problems, reordering modules, changing the sequencing of explanation and challenge — a new method emerged: problem first, explanation second. Give students a problem they have never seen. Let them struggle. Then, and only then, reveal the concept. The struggle is not the obstacle. The struggle is the learning.

The course that emerged from all of this recreates, deliberately, the condition Sandeep grew up in. Every concept is presented as a problem you have never been given the answer to. No explanation comes first. No algorithm is shown before you attempt to build one. You are placed, intentionally, in the position he occupied as a child in a village with no teacher: a problem in front of you, and no option but to figure it out.

You design. You build. You defend. And then — only after you have invented your own version — you discover what the world calls it and why it was designed that way. By the end, students don't just know algorithms. They own them. The knowledge was never handed to them. They found it themselves. And that is the only kind that lasts.