White-Label LLM Web Interface

I need a fully white-labelled large-language-model solution built around either Qwen or DeepSeek. The model must sit behind a clean, responsive web interface that feels product-ready to end consumers and can later be re-skinned with our own branding.

Core interface requirements
• Interactive text input/output so users can chat with the model in real time.
• Data visualisation panes where generated analytics or structured outputs from the LLM can be plotted or rendered as charts/graphs.
• A simple yet secure sign-up / sign-in flow (email-password or social OAuth) to keep conversations private and allow session resumption.

Technical expectations
The stack is flexible as long as it is modern and maintainable—Python (FastAPI, Django) or Node (Express, Next.js) for the backend, React / Vue / Svelte for the frontend, and Docker for one-command deployment are all acceptable. Whichever model you choose—Qwen or DeepSeek—please expose configuration files so I can swap checkpoints or adjust inference parameters without touching the codebase.

Deliverables
1. Source code for both frontend and backend, clearly structured and documented.
2. Model integration layer that loads either Qwen or DeepSeek with inference settings separated in a config file.
3. Auth module ready for production (hashed passwords, JWT or equivalent).
4. Visualisation component that takes model output in JSON or CSV form and renders charts through a library such as D3, Chart.js, or ECharts.
5. Deployment guide (Docker Compose or similar) plus a short README describing how to rebrand the UI.

Acceptance criteria
• User can create an account, log in, and start a chat within two clicks.
• Responses return in under three seconds on a GPU-equipped server.
• Switching from Qwen to DeepSeek is a one-line change in config.
• All assets (logos, colours, favicons) live in a single theme folder for quick white-labelling.