Girish Narayanswamy
Research Assistant at University of Washington
Seattle, Washington, United States
Summary
👤
Senior🎓
Top SchoolGirish Narayanswamy is a PhD student and research assistant at the UW Ubiquitous Computing Lab, with a decade of experience building ML and embedded systems for multimodal sensing and mobile health. Advised by Prof. Shwetak Patel and formerly a student researcher at Google, he works at the intersection of signal processing, learning for resource-constrained platforms, and health sensing across wearables, vision, audio, and radar. He has hands-on experience productionizing deep learning pipelines—evidenced by contributions to the NeurIPS 2023 rPPG-Toolbox where he fixed training and data-loading issues for remote PPG models. Prior roles at OctoML and Uplinq reflect his ability to bridge firmware, FPGA/accelerator toolchains, and efficient on-device inference. Known for combining rigorous research with practical engineering, he focuses on scalable time-series and health foundation models that run on constrained hardware.
10 years of coding experience
6 years of employment as a software developer
Doctor of Philosophy - PhD, Electrical and Computer Engineering, Doctor of Philosophy - PhD, Electrical and Computer Engineering at University of WashingtonBachelor’s Degree, Electrical Engineering, Computer Engineering, Bachelor’s Degree, Electrical Engineering, Computer Engineering at University of Colorado Boulder
Spanish, Tamil, English
Github Skills (11)
data-preprocessing10
data-loading10
pytorch10
computer-vision10
deep-learning10
dataprep10
trainings10
preprocessing10
python10
modeling10
preprocess10
Programming languages (3)
Github contributions (5)
ubicomplab/rPPG-Toolbox
Aug 2022 - Jan 2023
rPPG-Toolbox: Deep Remote PPG Toolbox (NeurIPS 2023)
Role in this project:
ML Engineer
ML EngineerContributions:25 reviews, 15 commits, 34 PRs in 4 months
Contributions summary:Girish primarily contributed to the development and maintenance of the rPPG-Toolbox, focusing on bug fixes, data loading, and training procedures. They modified the training loop within the EfficientPhysTrainer to correct an issue with chunk length and frame processing. They also addressed data loading issues and refactored dataset splitting for the PURE dataset, ensuring correct subject splits, and corrected naming conventions. Their changes indicate a strong understanding of the deep learning model training and the underlying data processing pipelines used for remote physiological measurement.
cameracardiovacomputer-visiondeep-learninghealth
girishvn/CU-CSCI3308-CIPractice
Apr 2017 - Mar 2023
Contributions:5 pushes in 6 years
continuouscontinuous-integration
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.