Why on-device?
Local inference avoids sending raw audio to a remote server.
RESEARCH / PRESENTATIONS / RECOGNITION
Questions worth
working on.
Applied research, shared ideas and lessons from taking a prototype into the real world.
Google Scholar (opens in a new tab)EMBEDDED MACHINE LEARNING
CoughSense
Arduino TinyML-based real-time cough detection.
A team prototype exploring cough detection entirely on an Arduino Nano 33 BLE Sense, keeping audio on the device. The workflow combines Coswara audio data, Mel Filterbank Energy features, a convolutional neural network and Edge Impulse deployment.
Authors: Adit Shah, Darsh Rank and Pratham Patel
Course: CSC 542 · NC State
Open full-resolution poster (PDF) What did we explore?
Audio features, augmentation and neural-network deployment on constrained hardware.
What’s next?
More investigation into robustness, evaluation and resource tradeoffs beyond the prototype.
PUBLICATIONS
Written & published.
Publication records and citation details are available on Google Scholar.
Virtual Reality & Intelligent Hardware
Advanced Driver Assistance System (ADAS) and Machine Learning (ML): The dynamic duo revolutionizing the automotive industry
Intelligent Strategies for ICT · Springer
Innovation in Agro-trade: Driving Forces Behind India’s Export and Import Growth
CONFERENCE PRESENTATION
CppCon 2026
Accelerating Quantitative Feature Engineering — A C++20/pybind11 Extension for a Python-Based Trading Research Platform
Part of AEQUITAS, my quantitative research platform.