Building tools so intelligence isn't gated by labs, geography, or money.
I research, design, and build at ArthaLabs — a one-person lab I founded to make AGI cheap enough to run at the edges of the world.
Open to research collaborationsAbout
I am a class 12 CBSE student in India, and I build ArthaLabs as an independent research lab focused on affordable AGI, model reliability, hidden-state probes, continual learning, and evidence-grounded AI systems.
I am trying to understand how smaller and more local systems can expose internal evidence, update knowledge without full retraining, and remain useful outside high-resource English and frontier-lab settings.
Proud of
Now
I am turning ArthaLabs into a focused research program for cheap, reliable, local intelligence: hidden-state readouts for false-pass detection, continual-learning memory systems, reservoir computing, and low-resource model training that does not assume frontier-lab budgets.
Most days are split between experiments, benchmark harnesses, careful writeups, and making the public evidence packet cleaner than the claims around it.