On Ground Labs
Intelligence for the real world
Started 2026
India's AI moment is here.
But most AI research lives in the clouds. Trillion-parameter models. Impossible datasets. Built for data centers, not for people.
I want to build the opposite.
THE HYPOTHESIS
AI research today is too abstract, too elite, too disconnected. A teacher in a small-town school can't use GPT-4. A student in tier-3 India can't experiment with models that need 80GB of VRAM.
Intelligence shouldn't require a data center.
WHAT I'M EXPLORING
Small Language Models — Compact models that run on affordable hardware. Built for India's context. A model that fits on a ₹15,000 device and still helps a student understand calculus.
Open Research — Papers designed to be reproducible. Codebases public and modular. Experiments documented for the next learner.
Research Mentorship — Training undergrads and early-career engineers to publish real papers. Co-authorship opportunities. Guidance from idea to publication.
THE APPROACH
I'm planning to run it as both a research lab and a learning platform.
Students and independent researchers could collaborate on publishable work. They'd contribute experiments, evaluations, fine-tuning. They'd get credited as co-authors.
The vision includes research mentorships for undergrads preparing for graduate studies. Helping them conduct real research. Build portfolios. Get personalized guidance.
Collaborating with universities, government programs, and research initiatives. Keeping all papers and code open forever.
THE PHILOSOPHY
"On Ground" means applied, field-tested, close to life. The opposite of "in the cloud."
The hypothesis: the best intelligence isn't theoretical. It's embodied in systems that work in everyday life.
Humility over hype. Experimentation over speculation. Usefulness over virality.
THE VISION
If this works, I'm hoping to:
Publish research in small model training, reinforcement learning, and applied AI.
Partner with schools and universities to create research opportunities for undergrads and early-career engineers.
Develop small-scale AI systems and open-source datasets for experimentation and education.
Build connections across India and beyond. Students, researchers, and practitioners around shared purpose.
Long-term: grow into an independent institute at the intersection of science, engineering, and social good.
THE DREAM
While others build models that live in data centers, I want to build systems that live in the world.
A chatbot that helps a student understand concepts.
An SLM that runs on affordable hardware and makes someone's day easier.
Research that a tier-3 college student can reproduce and learn from.
Not chasing fame. Building foundations.
Intelligence grounded in people, purpose, and the world.