Now

What I’m doing right now.

Updated September 2026 · Tamil Nadu, India

Building

Taking on selected client projects via f4ust — websites and web apps for founders and growing teams. I work alone and take each project from first scope through launch.

Also actively working on improvements to SprintUI and experimenting with new open-source tools when something useful comes up.

Studying AI

My current research focus: reasoning capacity in small open-weight language models. Specifically — how much reasoning headroom exists at inference time, why it’s systematically underestimated, and how to reliably unlock it without retraining.

I recently published my first paper on this: Adaptive Reasoning Has Measurable Headroom in Small Open-Weight Models. Working on follow-up experiments now.

Writing

Publishing research notes and essays on this site. Trying to write things I’d want to read — not summaries of other papers, but observations from building systems and running experiments.

Read the archive →

Reading

Frontier AI model papers, systems research, and occasionally things completely unrelated to code.

This page follows the /now convention. It’s a snapshot, not a live feed — updated when things change.