Meghna Roy Chowdhury

PhD Student, Electrical & Computer Engineering @ Purdue University

"Arise, Awake & Stop Not, till your goal is reached!"

Hello there! I'm Meghna, a PhD student in ECE at Purdue University, focusing my doctoral research on low-power wearable tech and AI in healthcare, while also bringing experience in applied ML for circuit optimization and hardware security.

My research involves system-level applications combining low-power sensing, communication, and analysis of various biopotential signals. I utilize techniques from Signal Processing, Machine Learning, Deep Learning, and Computer Vision, and also work on data collection and building predictive models.

Currently researching at SPARC Lab, I enjoy blending hardware and software skills to create impactful solutions.

To sum it up: I am a passionate, hands-on, and always-ready-to-learn-and-apply-new-tech kind of individual, seeking opportunities to contribute my best in the domains of healthcare AI, low-power wearables, circuits, and secure intelligent systems.

📍 Based in West Lafayette, IN (Boiler Up!) 🚂

🤯 Fascinated by low-power computing & clever algorithms

📷 Amateur Photographer

🛠️ Technical Toolkit

A collection of languages, tools, and concepts I work with.

Software and Coding

Python C / C++ Micropython Embedded C MATLAB Ansys HFSS Verilog HTML/CSS/JS

Hardware & Embedded

Microcontrollers (Arduino/NodeMCU/PSoC) FPGA (Artix-7) Raspberry Pi SDR (HackRF, USRP)

Subject Knowledge

Machine Learing Signal Processing Embedded Systems IoT & IoB Biopotential Signals (ECG/EMG/EEG/PPG)

Soft Skills (i know it's not a technical thing, but still)

Great Collaborater Problem Solver Adaptable Accountable Good Time Manager Analytical Thinker

⏳ Experience Timeline

My journey through research labs and industry roles.

May 2025 – Aug 2025

ML Intern

Qualcomm, Santa Clara, CA

Focus: Machine learning for circuit optimization.

Manager: Kamal Aggarwal

Aug 2021 – Present

Graduate Research Assistant

SPARCLAB, Purdue University

Focus: Health AI, low-power wearables, ML in hardware security.

Advisor : Shreyas Sen

Aug 2020 – Aug 2021

Developer - R&I (Analytics & Insights)

TATA CONSULTANCY SERVICES (TCS)

Developed ML solutions for structured & unstructured data.

May 2020 - Jul 2020

Research Intern

YUAN ZE UNIVERSITY, TAIWAN

Focus: Signal processing & time-series analysis using CNNs/LSTMs for ECG and PPG.

Advisor : Prof. Jiann-Shin Shieh.

Jan 2020 – Apr 2020

Machine Learning Intern

TATA CONSULTANCY SERVICES (TCS)

Focus: Predictive modeling for IoT applications in EV battery life prediction and range estimation.

May 2019 – Jun 2019

Research Intern (SEAL Lab, CSE)

IIT KHARAGPUR

Focus: Hardware security primitives, FPGA implementation and testing.

Advisor: Prof. Rajat Subhra Chakraborty.

May 2018 – Jul 2018

Summer Intern (EE Department)

IIT BOMBAY

Focus: Embedded programming, sensor interfacing, cloud communication.

Guidance: Prof. Virendra Singh.

👩‍💻Projects and Publications

Featured Projects

Bringing ideas to life with code and hardware. More on GitHub.

I-HOPE Project Visual

I-HOPE

Predicting college student mental health using interpretable AI to understand complex behavioral data.

I-HOPE Project Visual

DistNN

Enabling powerful, real-time AI on wearables with extreme energy efficiency (~1000x > GPU) via distributed neural networks.

I-HOPE Project Visual

SSL-SE-EEG

Achieving state-of-the-art EEG analysis accuracy (91% on MindBigData) with less labeled data using a novel Self-Supervised Learning framework.

I-HOPE Project Visual

FlexECG

An ultra-low power (20µW) flexible ECG patch utilizing body heat and Human Body Communication for seamless monitoring.

IDD Simulation Results

Precise Drug Delivery

A novel electronic system using electric fields for accurate, potentially lower-cost Intrathecal Drug Delivery (IDD).

NeuroType: EMG-Based Typing via Finger Classification & Transformer Decoding

The Challenge: Typing Without Touching, For Anyone

Using muscle signals (sEMG) for silent, hands-free typing is a promising goal. However, EMG signals are noisy and vary hugely between users, making robust, user-independent systems a major challenge.

Our Solution: NeuroType - A Modular Pipeline on Large-Scale Data

We developed NeuroType, a system translating raw wrist sEMG into sentences without user-specific training. Crucially, this work leverages the large-scale `emg2qwerty` dataset released by Meta Reality Labs, featuring over 346 hours of typing data from 108 users. NeuroType mimics typing:

Predict Fingers: Identifies finger activations using a BiLSTM+Attention model.
Narrow Down Letters: Uses keyboard ergonomics to create likely letter pools.
Generate Sentences: A Transformer (T5) language model constructs coherent sentences from the possibilities.

This structured approach enables robust zero-shot generalization across users.

Impact & Results: 🚀⌨️

Achieved highly accurate sentence prediction: 1.6% Character Error Rate (CER) and 1.8% Word Error Rate (WER).
Outperformed previous personalized baselines by over 4x – without needing calibration for new users!
Performed the first detailed analysis showing structured EMG patterns related to finger use in the extensive `emg2qwerty` dataset.
Advances practical, silent, hands-free text input for wearables, AR/VR, and assistive tech.

Tech Highlights: Surface EMG (sEMG), BiLSTM-Attention Networks, Transformers (T5), Ergonomic Priors, Zero-Shot Generalization, Python (TensorFlow, Hugging Face), `emg2qwerty` Dataset.

View Publication (UbiComp/IMWUT)

Publications

Here are some of my published research findings. More on Google Scholar.

Chowdhury, M. R., Xuan, W., Sen, S., Zhao, Y. & Ding, Y. (2025). 'Predicting and Understanding College Student Mental Health with Interpretable Machine Learning'. In IEEE/ACM Conference on Connected Health: Applications, Systems and Engineering Technologies (CHASE). [Link]
Chowdhury, M. R., & Sen, S. (2025). 'Measurement and Analysis of System Parameter Effects on Noise in EEG Systems'. In IEEE International Symposium on Circuits and Systems (ISCAS). [Link]
Chowdhury, M. R., Ghosh, A., Bari, M. F., & Sen, S. (2024). 'Leveraging Ultra-Low-Power Wearables Using Distributed Neural Networks'. In IEEE Body Sensor Networks Conference (BSN). [Link]
Bari, M. F., Chowdhury, M. R., & Sen, S. (2023). Long Range Detection of Emanation from HDMI Cables Using CNN and Transfer Learning. In Design, Automation & Test in Europe Conference & Exhibition (DATE) 2023. [Link]
Bari, M. F., Chowdhury, M. R., Chatterjee, B., & Sen, S. (2022). Detection of Rogue Devices using Unintended Near and Far-field Emanations with Spectral and Temporal Signatures. In IEEE MTT-S International Microwave Symposium (IMS) 2022. [Link]
Chowdhury, M. R., Madanu, R., Abbod, M. F., Fan, S. Z., & Shieh, J. S. (2021). Deep learning via ECG and PPG signals for prediction of depth of anesthesia. Biomedical Signal Processing and Control, 68, 102663. [Link]

🎓 Education

My academic journey so far.

2014

Grade X (ICSE)

Ebenezer Intl. School
Bangalore, India
Score: 96%
2016

Grade XII (CBSE)

Jaipuriar School
Mumbai, India
Score: 92%
2016 - 2020

B.Tech, ECE

Vellore Institute of Technology
Electronics & Communication Engineering
India
CGPA: 9.24 / 10.0
2021 - Present

PhD, ECE

Purdue University
Electrical & Computer Eng.
West Lafayette, IN

💪 Beyond the Lab

When I step away from the screen and soldering iron...

📷Photography

🏊Swimming

🚶Hiking

🧗Bouldering

🏸Badminton

✈️Travel Bug

A Glimpse Through My Lens

More on Instagram. Click photos to enlarge.

Heart shape in sky

Seagull eating

Sunset silhouette

Street motion blur

Navi Mumbai skyline

Crescent moon

Flamingos

Street scene reflection

Howrah Bridge

Landscape view

Myna bird

Painted Stork bird

🤝 Get In Touch

Whether you have questions, ideas for collaboration in low-power tech or AI in healthcare, or just want to chat about research (or photography!), please feel free to reach out!

Email: royc.meghna@gmail.com

Or schedule a quick virtual coffee chat:

Schedule a 30-min Call