I need a fully working prototype that detects five organ-specific diseases—liver, kidney, gastro, lungs and intestine—using a transformer architecture. I will share Kaggle dataset links for each category the moment we start, so there is no data-gathering lag. Model • Train a single or ensemble transformer that classifies images into the five target organs. • Provide a clear training script and an exportable, reproducible checkpoint. • Include concise inference code so I can run a quick test locally. API Expose a lightweight endpoint (FastAPI or a similarly fast Python framework is fine) that the front end can hit for predictions. A simple Dockerfile would simplify hand-off. Frontend • React with a clean, minimalist look—no heavy styling frameworks beyond something like Tailwind or plain CSS. • Users should be able to drag-and-drop or browse to an image, press “Predict,” and instantly see the top class plus confidence scores. Delivery timeline Everything—model, API and React UI—needs to land in my inbox by tomorrow. Please push to a Git repo so I can clone, install, and run start-to-finish with a single command. Acceptance I should be able to: 1. `git clone`, `docker compose up` (or similar) 2. open `localhost:3000`, upload a test image 3. receive a correct class label returned by your model. If you can commit to this timeline and deliver the pieces exactly as described, let’s start immediately.