Browser agents need better eyes
A talk and working thesis about browser agents, observation design, and why better models still fail when the state is bad.
Things
Things I have made, studied, or keep returning to.
A loose shelf for projects, talks, research, experiments, and unfinished threads. Some are public artifacts. Some are just the shape of a question I keep pulling on.
Production voice-agent systems at Sarvam
Real-time voice-agent and LLM systems for Indian-language phone, WhatsApp, and enterprise contexts.
Vision-guided caption correction for WAT 2025
A public ACL paper on using visual context to find and correct caption errors in English-to-Indic multimodal translation data.
Diffusion models for medical imaging
IIT Madras thesis work around diffusion models, medical imaging, and the data problems that appear before models can be useful.