Available for work
Dhruv Kotegaonkar — Full-stack engineer, real-time AI systems. Currently in India.
Voice agent shipping soon — you'll be able to talk to this page
Selected work
A real-time conversational agent embedded in this page. Speak to it and it answers as me — streaming audio both directions over a single WebSocket, with barge-in so you can interrupt mid-sentence.
Sub-second time-to-first-audio by pre-warming the model session before the connection opens.
Telephony-grade AI calling: bidirectional μ-law audio bridged to a live multimodal model, with interruption handling, session resumption, transcript capture and post-call analysis.
TODO: add a concrete number — call volume, latency, or accuracy.
Pick something you can demo live and talk about in depth. Depth on three projects reads far better than nine thin ones.
TODO
About
I work on the unglamorous parts of real-time AI: audio transcoding, interruption handling, session lifecycles, and everything that has to hold together before a voice agent feels like a conversation rather than a demo.
TODO: how you got here — what you studied, what pulled you toward this work.
TODO: what you're looking for next.
Toolkit
Open to full-time roles and interesting problems in real-time AI. The fastest way to reach me is the button below — or just talk to the agent on this page.