I am the Director of Algorithms at Eko Health. Previously, I have worked at Graphcore and iRhythm Technologies. My work interests include statistical/machine learning, deep learning, digital health, autonomous systems, robotic and mobile sensor networks. During my academic life, I was an NSF IGERT Fellow at UC, Santa Barbara and made contributions to the theory of fluid mixing and ergodic control.

Google Scholar Profile

LinkedIn Profile

Publications

My recent work with Text-To-Speech and Speech Recognition Models:

  1. A short blog post on generating speech from text with Graphcore IPUs

  2. Code base for Deep Voice 3 implementation on Graphcore IPUs

  3. Speech Recognition using Conformer model with Connectionist Temporal Classification (CTC) training

  4. Speech Recognition with Transformer-Transducer and with RNN-T loss

Some links relevant to my academic work:

  1. Code repository for coverage algorithms for multi-agent systems - Spectral Multiscale Coverage (smc).

  2. Review article on multiscale metrics (mix-norms) to quantify fluid mixing -  a concept I developed for my thesis.