From web developer to an applied-AI studio: how Bodham Labs started

Where did it start?

With websites. From 2015 to 2019 I was a full-stack web developer at Techxilla and D2WEB, working in PHP, WordPress, Shopify and JavaScript. Between 2019 and 2022 I did my MCA at Himalayan Garhwal University. From September 2022 to December 2023 I was a senior web developer at Sago Payment Solutions, working on a payment gateway and Stripe integration.

Two habits from those years still shape our work. Web clients judge a site by whether it works on Monday morning, not by how it is built. And payment work teaches you that every request needs a failure path, because a half-finished payment is worse than none.

Where does teaching come in?

I hold a BTech in Electrical Engineering, an MCA and a B.Ed. My wife and I run RK Institute of Learning in Delhi, and she co-runs the teaching. I teach Class 11 and 12 academics, and job-readiness skills to BCA students.

Teaching changes how you write. A student who does not follow the first sentence will not read the third. That is why our notes give the short answer first, use short sentences, and ask headings as questions.

What did we build in 2026?

I started building AI systems seriously in early 2026. The work followed one thread: what does it take for AI to be useful every day, on ordinary hardware? Each row below has its own note.

Some of it shipped. Version 4.0.0 of NeoGravity, a small tool that keeps a coding assistant's context between sessions, went on npm in March 2026. Some of it failed, and those failures taught us the most.

How do BCA students work with us?

As apprentices. We mentor BCA undergraduates and technical interns as junior developers, automation operators and teaching assistants. They work on real tasks under review, not on toy projects.

This is the other half of the teaching. Real systems fail in ways no syllabus covers, and students learn most from seeing why.

What does Bodham Labs do now?

Bodham Labs is a Udyam-registered MSME, founded in 2026 in Delhi. We work for businesses in India and abroad, and for students.

• Services: operations automation, private AI search, agent infrastructure, Hindi and Hinglish AI, and data extraction.
• Training: AI, ML and agents, for teams and students.
• Bodham: our learning platform for school students. Status: launching soon.

Why do we publish our dead ends?

Because a studio that only shows wins is asking you to trust it. We would rather show the work. Every project in our work ledger carries a grade: in use, measured, prototype or exploring. A number appears only with the setup it was measured on, or with “not measured yet”.

So the notes include the model that fired nearly 17,000 order requests in paper mode, the hidden state that reached 10¹⁵, and our own review that called an agent “over-engineered”. When a test is too small, we say so and re-run it before publishing a figure.

What this does not prove

• A story is not evidence. Judge us by the ledger and the notes, each with its setup.
• Statuses change. Every status here is as of October 2026.
• Most 2026 work was research. Of the projects here, only RK Ops runs a real operation every day.

Questions