Mbughuni Mvungi

A horticultural export company needed a production-grade web presence to represent its fine beans, snow peas, and chili exports, with performance that holds up for real customers.

A full-stack platform with a React/Next.js frontend and a Node.js + Supabase backend, deployed on Render.

• In-memory API response caching with TTL-based invalidation on product routes
• Image performance optimization via lazy loading and Supabase image transform parameters
• Brand/UI system implementation across the site

Access to reliable, understandable health information is limited for many users — they need a way to ask health questions and get educational guidance, not a diagnosis.

An AI-powered healthcare assistant backed by large language models, with a FastAPI backend handling query processing.

• RESTful APIs built with FastAPI
• AI model integration for processing user health-related queries
• Focus on educational guidance and improving access to health information

Time-series forecasting is a core ML skill — this project applies it to a real, verifiable problem: predicting Apple (AAPL) stock price movement.

A Long Short-Term Memory (LSTM) neural network trained on historical AAPL price data.

• Implemented in Python with TensorFlow, NumPy, and Pandas
• Model evaluated using standard regression metrics
• End-to-end pipeline: data preprocessing → training → evaluation

A publisher needs a full-stack platform to manage and publish articles with an editorial workflow, not just a static blog.

A full-stack web application with a React.js frontend, Node.js backend, and PostgreSQL persistence.

• Admin panel for content management
• Article management workflow
• Dynamic content rendering

Renting or selling property needs a mobile-first experience with real-time listing updates and secure user accounts.

A Flutter-based mobile application with Firebase for authentication and real-time data.

• Firebase authentication
• Real-time database for live listing updates
• Property listing, search, and management