Eyad-Elghonemy▊

# Production-grade Titanic survival prediction API built with FastAPI and TensorFlow/Keras. Takes raw passenger data (age, fare, class, family info) and returns survival predictions using a trained neural network, with support for single and batch classification.

# Full-stack deep learning service for Fashion-MNIST clothing classification. FastAPI backend serves a trained Keras model via a secured REST API with Supabase-backed logging, paired with a Streamlit control room for single/batch classification and live usage analytics.

# Production-ready Arabic sentiment analysis API built with FastAPI and scikit-learn. Ships with a pre-trained model (~79% accuracy on 58K Arabic tweets) and supports live retraining on custom data with zero downtime — swap models in and out via simple API calls.

# Predicts bank customer churn using XGBoost, served via FastAPI and visualized with Streamlit.

# AI-powered vehicle price evaluation API for the Egyptian insurance market. FastAPI backend with VIN-based prediction, full ML pipeline, and Supabase-backed logging.

# End-to-end machine learning pipeline for heart disease risk prediction using the UCI dataset. Covers preprocessing, PCA, feature selection, supervised/unsupervised learning, and hyperparameter tuning, deployed as an interactive Streamlit dashboard with real-time predictions.