MD. FAHIMUL ISLAM

A robust Python crawler built with Selenium and BeautifulSoup that scraped and organized 15,339 U.S. municipality images across all 50 states.

Regression modeling that predicts state-level student scores from demographic and financial indicators to support data-driven education strategy.

A Python machine-learning model that classifies spinal abnormalities from biomechanical angle measurements to aid clinical decision-making.

Automated demand forecasting built for reproducibility — Azure ML training and registry, Azure DevOps CI/CD, containerized deployment to ACI and Kubernetes, and drift-monitored retraining.

A local-first GenAI/RAG platform that turns dense clinical-trial protocols into searchable, cited answers — built to keep sensitive data inside a private environment.

A TensorFlow/Keras deep-learning model that predicts rheumatoid-arthritis disease activity from clinical features to support personalized treatment planning.

A predictive model for U.S. high-school dropout paired with an interactive Plotly dashboard that surfaces the biggest contributing factors for policymakers.

The robots, modules, and systems projects that shaped my engineering — competition robots that read sheet music, a Yocto-based Linux build for the STM32F429, and years of R&D hardware.

Two 2026 research directions bringing deep learning to wearable health — detecting hypoglycemia from smartwatch heart-rate dynamics, and privacy-preserving TinyML federated learning on low-power IoT.

Building trustworthy machine learning in production — a private clinical-trial RAG platform, an Azure ML forecasting pipeline with drift monitoring, and medical diagnostic models.