Our mission
Every person deserves a
shot at this knowledge.
Not the person who can afford a $2,000 bootcamp. Not the person whose native language happens to be English. Every person — the one switching careers, the student in a country where the dollar is expensive, the builder who already has a job and steals two hours at night.
Why this exists
I started out studying biology. Not software, not math — biology. Cells, systems, the mechanics of living things. I liked how it all connected, but I had no idea that the same instinct — wanting to understand how systems work — would eventually drag me into computers.
The jump wasn't planned. I taught myself to code the way most people do: scattered YouTube videos, half-finished tutorials, forum threads that assumed you already knew three things you'd never heard of. There was a lot of watching and nodding and feeling like you understood, and then sitting in front of a blank file and realising you didn't. I got there eventually — but I lost a huge amount of time to content that wasn't designed for someone who needed to actually build something.
Mathematics came third, through necessity. Once I got into machine learning and then cryptography, there was no way around it. The math was no longer abstract — it was the thing that explained why the code behaved the way it did. Learning it that way, backwards from a concrete problem, was completely different from how I'd been taught in school. It clicked. It stuck. And I started to wonder why that approach wasn't how most technical education was structured.
That's the question Capstok is built around. Not what should someone learn, but how does learning actually work for builders — and can we build a platform that costs almost nothing and still does it right?
One more thing worth being direct about: these aren't courses I built because I've mastered the material. They're the courses I'm working through myself. Capstok is me sharing my learning journey in public — the order I tackled things, the explanations that actually clicked, the projects that forced ideas to land. If you're somewhere behind me on the same path, that's exactly who this is for.
— Lakshya Kumar, founder
How we operate
Minimal price. Always.
Pricing is a values decision. We keep memberships low enough that a student, a career-switcher, or anyone in a lower-income market can afford to join without thinking hard about it. The first module of every course is free with no sign-up. You can verify your email and unlock more for nothing. If you want the full catalog, it's a few dollars a week — not a few hundred dollars a month.
No stale content.
We don't record videos and hope the world stays still. The curriculum is text — which means we can update a lesson the day after a spec changes, a library drops a breaking version, or a better explanation exists. You will never land on a lesson that references a tool from three years ago as if it were current. Outdated content is a bug we treat as urgently as a broken feature.
No passive watching.
You know the feeling. You're fifteen minutes into a video, you space out for thirty seconds, and you've lost the thread. You rewind. You watch again. You nod. You close the tab and open a blank file and nothing is there. We don't do videos because we don't believe they work for the kind of learning we care about. Every lesson on Capstok is something you do — a short challenge, a prompt to run, something to break, something to build — not something you sit back and absorb.
Search, look, and do.
We want you looking things up while you work. The lessons point at primary sources — specs, papers, canonical docs — not summaries of summaries. You have an AI next to every challenge that knows the curriculum and won't hallucinate the answer because it has the actual content in context. The goal is to build the habit of finding out, not the habit of remembering. That habit is what makes a builder.
What we're building toward
AI isn't a field. It's a layer on every field.
The doctor who understands how a diagnostic model actually works will use it better than the one who treats it as a black box. The lawyer who knows what a language model can and cannot do will write better prompts than one who just pastes in a question. The designer, the analyst, the teacher, the researcher — every field has an AI layer forming on top of it right now. Most people in those fields aren't learning it, because the content that exists was written for software people.
We're building courses for the rest of them. Not AI for beginners — that exists, it's fine, you can find it for free. We mean courses that meet you inside your existing field and show you what AI changes there specifically: the workflows, the failure modes, the prompting patterns, the things that will make you faster and the things that will trip you up if you trust the output blindly.
AI-assisted diagnostics, literature review, clinical decision tools
Contract analysis, case research, regulatory interpretation
Model risk, quant workflows, AI in trading and reporting
Curriculum design, adaptive tutoring, assessment generation
Generative tools, taste vs. automation, the human edit
Literature synthesis, hypothesis generation, reproducibility
These courses are in development. If you work in one of these fields and want to help shape the curriculum — reach out. We want to build with people who are already inside the problem, not just observe it from outside.
The commitment
A skilled, educated generation — regardless of language or budget.
That's the sentence everything above is in service of. The pricing, the format, the no-videos rule, the constant updates — they're all downstream of that one thing. If you're here to build, we want to be the place that meets you where you are.