Moksh Bainsla

Reform yourself, and you reform thousands.

This is a working notebook, left open. What I'm learning, building and getting wrong — shared in case some of it is useful to you.

Each ring is one person changing, then passing it on. Tap anywhere to drop another stone.

The idea

I don't think change starts with a plan for the world. I think it starts with one person getting a little more honest, a little more skilled and a little more careful with other people — and that showing up in their work.

Good work travels further than any speech. A clear explanation gets passed on. A tool that genuinely helps gets used by people you'll never meet. A teacher who cared becomes the reason someone else starts teaching.

So the order matters. Work on the centre first, and let the outer rings take care of themselves.

Nothing on this site is a finished answer. It's the inner ring — the part I can actually work on — written down so you can see the process, not just the result.

Four rings, from the centre outward The centre is you. Around it, your work. Around that, the people your work reaches. On the outside, the people they reach. You your work the people it reaches the people they reach

What I'm working on

Not a list of wins. Each of these is something I'm still in the middle of, with the one thing it's teaching me that you could use too.

  • Learning AI from the ground up

    Ongoing

    Working through computer vision and deep learning properly, as notebooks where every diagram is live — drag it, slide it, break it. Alongside that, a connected map of how maths, AI, systems, physics and society link to one another.

    What you can take

    Before reading how something works, ask what problem forced someone to invent it. The answer usually explains the rest.

  • Manu, a personal AI agent

    Early

    Building an agent from scratch, one capability at a time — a brain, then memory, then tools, then planning. Each step has to work and be understood before the next one starts.

    What you can take

    Discover, build, test, break, debug, understand — then level up. Turning learning into small quests makes it hard to fake.

  • Trinetra

    Most of the way there

    A platform for searching and working with satellite imagery. Most of the system is built; finishing it properly is the work now.

    What you can take

    Building is the exciting 90%. The last 10% is quiet, unglamorous and the part that actually ships. Plan for it.

  • Kahani

    Working prototype

    A tool for carrying stories across Indian languages, so something told well in one language can reach people who think in another.

    What you can take

    Language is one of the quietest barriers to learning. Removing it is often worth more than adding new content.

  • Finding teachers who inspire

    Day job

    At Masters' Union I look for practitioners who can teach AI and data to students — people who know their craft and make others want to learn it. Before this, four years at upGrad running more than twenty AI and data programmes for working professionals.

    What you can take

    People rarely remember the syllabus. They remember the person who made them believe they could do it.

  • Sikho and Disha

    Prototypes

    Early learning and guidance tools for students the system tends to overlook — no pedigree, no network, no one to ask.

    What you can take

    Content is almost free now. What's still scarce is guidance and someone who believes in you. Be that for one person.

Notes I keep coming back to

Small rules I've arrived at by getting things wrong first. They're not original — just tested.

Go to the bottom of it

What is this really? Why does it work? What's the simplest version? Understanding beats memorising every time.

Believe what runs

Doubt what's only pitched. A small thing that works teaches more than a big plan that doesn't exist yet.

Build it small, then break it

Write the thirty-line version first. Break it on purpose. Then read how the professionals did it.

Automate the predictable

Never automate trust. The parts of work that are about people should stay with people.

Proof over pedigree

What you can show matters more than where you came from. That's good news for most of us.

Art and science aren't opposites

Rigor without imagination gets stuck. Imagination without rigor drifts. Most good work needs both.

Start with one ring

You don't need permission, a platform or a plan for the world. This is the whole method, as far as I can tell.

  1. Pick one thing

    Choose one thing you want to understand properly this month. Not ten. One.

  2. Make the small version

    Build, write or draw the simplest version that actually works. Let it be rough.

  3. Pass it on

    Explain it to one person. Teaching it will show you what you didn't understand — and give it to someone else.

Then do it again. That's how one ring becomes thousands.

This site changes as I do

A short record of what changed and when, so it stays honest.

  1. First version. The idea, the work in progress, the notes, and a way to start.