AI Medical Image Analysis.
AI Medical Image Analysis is a full-stack ml platform built by Yousef Ali Mansour. AI inference on chest X-rays and skin lesions across 12 condition classes, returning confidence-scored predictions.
At a glance
| Project | AI Medical Image Analysis |
|---|---|
| Type | Full-stack ML Platform |
| Built by | Yousef Ali Mansour — Founder & CEO at NUMU · Backend & Platform Engineer |
| Stack | FastAPI, React, PyTorch, PostgreSQL, Docker |
| Source | github.com/yousefalimansour/Ai-image-medical-helper |
| Last updated |
What I built
- Built a full-stack medical-imaging platform (FastAPI + React/TypeScript) running AI inference on chest X-rays and skin lesions across 12 condition classes, returning confidence-scored predictions to assist clinicians.
- Integrated a Vision Transformer (chest X-ray, via Hugging Face Inference API) and a ConvNeXt model (skin lesion, served locally as TorchScript) behind 5 versioned REST endpoints.
- Engineered with Clean Architecture (SQLAlchemy 2.0, Alembic, PostgreSQL), JWT auth, bcrypt hashing, and audit logging; fully containerized with Docker Compose.
Stack
FastAPIReactPyTorchPostgreSQLDocker
View on GitHub More projects
- NUMUMulti-tenant E-commerce Platform — Storefront engine, theme marketplace and integrated payments for Egypt & MENA — 15 live storefronts, 500+ orders.
- NeuroShieldAI Deepfake Detection · Graduation Project — Microservices platform that analyzes video to detect deepfakes and verify authenticity — Django + FastAPI inference pipeline.
- KnowledgeOps AIFull-Stack RAG Knowledge Platform — RAG platform with hybrid retrieval (Chroma + RRF + LLM rerank) and cited, streaming answers via Gemini 2.5 Pro — built to refuse, not hallucinate.
- Real-Time Stock AlertsDjango REST + Celery — Monitors 10 tech stocks and fires automated email alerts with distributed Celery + Redis scheduling.
- Automated Toll Collection API.NET 9 · Clean Architecture — 4-layer toll-collection REST API with EF Core, RS256 JWT auth and role-based access control.