Machine Learning · 2025

EHR – AI-Enabled Electronic Health Records Platform

EHR is an AI-enabled health record system designed for charity clinics and outreach programs. It streamlines patient registration and monitoring through a Streamlit-based interface backed by Flask and MySQL. Integrated modules built with Scikit-learn, TensorFlow, and Hugging Face provide health risk alerts, identity extraction, and automated reporting. The platform supports multilingual staff, operates reliably with offline-sync in low-connectivity settings, and helps clinicians deliver more efficient and accessible care.

Show more

Machine Learning · 2024/2022

FER – Facial Expression Recognition for Teletherapy

FER is a machine learning system designed to support psychological consultations by recognizing patient emotions during online therapy. Built for real-time environments using facial landmark detection and TensorFlow-based emotion models, the system enhances remote mental health care by helping clinicians observe subtle emotional cues. The platform uses WebRTC for live video processing, ensures privacy with consent gating, and seamlessly integrates into virtual therapy workflows.

Show more

Machine Learning · 2024/2022

BPM – Contactless Heart Rate Monitoring for Elder Care

BPM is a real-time, contact-free system designed to estimate heart rate by analyzing subtle facial landmark movements in live video. Built for deployment in retirement homes, the system replaces wearable sensors with a passive, non-intrusive monitoring layer that enhances safety during remote supervision. It combines facial tracking, motion filtering, and signal extraction to detect cardiac anomalies, and streams BPM data through WebSocket in less than a second, offering timely alerts to caregivers when every second counts.

Show more

Machine Learning · 2025/2023

HPO – Interactive AI Learning Portal

HPO is an interactive, browser-based platform that helps students understand how deep learning works through experimentation. Designed for hands-on education, it lets users upload datasets, configure neural networks, and visualize the impact of hyperparameters in real time. Built with TensorFlow, Docker, and Kubernetes, the tool delivers responsive model training and tuning experiences, even in resource-limited environments. Now used in live classroom settings, it lowers the barrier to ML literacy and bridges the gap between theory and practice.

Show more