Hello there!
I’m Shreekantha (Shree). I work on speech recognition and voice agents.
I spent nearly seven years at Dialpad, where I led the next-gen ASR effort that took the product from Kaldi to end-to-end systems, and was later the DRI for the cascaded voice AI agent architecture.
I came to speech through a Bachelor’s in Telecommunication Engineering at VTU and a couple of years at Sonus Networks, now Ribbon Communications, working on element management systems for 4G VoIP. I then did an MS by Research at IIIT-Bangalore, with a thesis on “Multi-task learning in end-to-end attention-based automatic speech recognition”.
What I work on:
- Streaming end-to-end ASR for conversational, telephony, and videoconferencing speech
- Realtime voice AI agents
- Low-latency and computationally constrained inference, increasingly on-device
- Multi-lingual and code-switched speech recognition
- Voice agents that use more than the transcript
I regularly work with K2, NeMo, Triton, ESPnet, LiteRT, PyTorch, and LiveKit.
Outside of work you’ll find me tinkering with:
- ASR on mobile NPUs
- AOSP
- On-device AI for LineageOS
- tinyML boards
You can reach me at nadig.shreekantha@gmail.com