AI Chat Platform — React Frontend

Frontend for the Multi-Model AI Chat Application — a ChatGPT-style chat app powered by open-source LLMs (Ollama). Users register, log in, create chat sessions, and send messages to models like Llama, Mistral, and Gemma.

The system is split across multiple repositories; each service is independent, containerized, and designed to scale. This repo delivers the React UI served by Vite in dev and Nginx in production.

• React 19 + Vite + React Router: Login, Register, Chat, Admin (/models) routes.
• JWT stored in localStorage; Authorization: Bearer on every API call.
• Async UX: POST message → receive task_id → poll until status is completed or failed.
• Admin page lists Ollama models when is_admin is true in login response.
• Dockerfile + nginx.conf for production; GitHub Actions CI/CD; VITE_API_URL=/api for K8s ingress.
• Full stack tech: Nginx reverse proxy, FastAPI gateway, RabbitMQ, Redis, PostgreSQL, Ollama, K8s on Grid'5000.

Chat UI

Sidebar session list, main chat area, protected routes, and 1.5s polling loop on GET /tasks/{task_id}/status.