About me

I'm Adarsh. I lead applied AI at pWin.ai, where I design and build AI agents and products for proposal and sales teams.

I've spent 13 years taking technical products to market. At Microsoft I helped launch Azure's first speech-to-text API and was a founding PM on Microsoft Stream. At Meta I worked on video infrastructure, including rolling out HDR video across Facebook, Instagram, and more.

Outside of work I build things and write about what I learn. Lately that means small language models, agents, and what it actually takes to make an AI system useful. Usually it's less about the model and more about everything around it.

Some recent projects: training a 4B model on a synthetic hospital benchmark, an iOS app that estimates drinks from a photo, and fine-tuning a speech model to grade my Mandarin tones (try it).

I also make music.