NAGHAM AMER

I'm building end-to-end AI applications across computer vision, RAG, time-series forecasting, APIs, and web interfaces.

In MediGuard Vision, I built a multi-stage medicine-inspection prototype combining YOLO detection, fine-tuned ResNet18 embeddings, ORB checks, and explainable MYCIN certainty-factor decisions. I delivered it as a FastAPI and React/Vite application with SQLite history, alerts, 60 backend tests, and a successful production frontend build.

For SupplySight AI, I prepared a leakage-safe demand-forecasting workflow from 1,067,371 retail rows, evaluated 4,555 products, and selected 541 forecast-ready products. I implemented TensorFlow and PyTorch training paths, MLflow tracking, Prometheus metrics, inventory recommendations, Docker Compose, automated API tests, and GitHub Actions CI.

I also built a context-grounded webpage RAG chat using LangChain, ChromaDB, Hugging Face, and TinyLlama, plus a Laravel and MySQL Property Finder with filtering, favorites, ratings, map-based search, admin approval, and secured APIs.