Projects Portfolio
AI-Powered Recipe Generator
Built an AI-driven web app using a fine-tuned GPT-2 model to generate custom recipes from user-provided ingredients. Deployed on Streamlit Community Cloud with Docker support for reproducible environments. Features robust output filtering to ensure coherent recipes and a modern, user-friendly interface.
HR Workforce Dynamics Dashboard
An interactive Streamlit dashboard for HR analytics, visualizing employee data such as salary distribution, growth trends, and turnover rates. Features a light blue/navy theme, animated GIF, and responsive charts built with Plotly. Cleans and processes PL/SQL-exported CSVs for actionable insights.
Laptop E-Commerce App – Built with Flutter
A cross-platform mobile app developed using Flutter and Dart for showcasing and selling laptops. The app connects to a REST API to dynamically load product listings, categories, and specifications. Key features include category filtering, wishlist management, and a functional shopping cart—all wrapped in a responsive, intuitive UI.
EEG-Based Sleep Stage Prediction
• Developed a Python-based system to classify sleep stages using EEG and physiological signals.
• Sleep recordings were segmented into 30-second windows, each labeled as one of five stages: W (wake), N1, N2, N3, and R (REM).
• Leveraged spectral patterns and transient EEG events (e.g., sleep spindles, k-complexes, slow waves) to enhance classification accuracy.
Arabic Twitter Sentiment Analysis
• Build a sentiment analysis system for Arabic tweets.
• Retrieved tweets using Tweepy and performed preprocessing: cleaning, tokenization, and lemmatization.
• Extracted features with TF-IDF and bag-of-words models to represent tweet semantics.
• Trained and evaluated classifiers (Naive Bayes & Logistic Regression) using scikit-learn for polarity classification (positive, negative, neutral).
E-commerce Cosmetic Store Recommender System
• Built a recommender system for big data (8M+ rows) using purchasing behavior data from an e-commerce cosmetic store.
• Utilized the Apriori algorithm and collaborative filtering to generate personalized brand recommendations.