Contact person: AI Render - low-quality CAD renders into photorealistic Web App

Phone:Show

Email:Show

Location: Wroclaw, Poland

Budget: Recommended by industry experts

Time to start: As soon as possible

Project description:
"I’m building a SaaS platform that converts low-quality CAD renders into photorealistic imagery through a custom AI model. The paired dataset (noisy vs. high-end outputs) is already prepared, and I have a GPU workstation where all model training and fine-tuning must run locally. Your task is to create the complete web application that lets end-users upload their render, queue it for processing, and download the finished visual while we handle recurring billing.

Core workflow
• Account creation with both email/password and the usual social logins.
• Single, recurring monthly subscription collected via credit or debit card (Stripe is the logical fit, but I’m open to alternatives that keep PCI scope minimal).
• Secure hand-off of uploads to my workstation for inference; the training pipeline and future retraining cycles also stay on that machine, so the web layer needs a clean API to trigger jobs, stream progress, and fetch results.
• Simple dashboard where users can view remaining render credits, past jobs, and payment status.
• Admin panel so I can retrain, monitor GPU usage, and manage subscriptions.

Key deliverables
1. Responsive front end (React, Vue or similar) with drag-and-drop upload and real-time status updates.
2. Back-end service (Node, Django, or FastAPI) that handles auth, Stripe webhooks, job queuing, and secure connection to my workstation (SSH tunnel or VPN-protected REST/gRPC).
3. Training/inference scripts wrapped so they can be called from the back-end without interrupting running sessions.
4. Deployment documentation that lets me stand the whole stack up on my server and keep the deep-learning pieces on-prem.

Acceptance is complete when a user can:
• Register (email or social), subscribe, upload a file, watch progress, and download the enhanced render.
• My admin view shows subscription status and GPU workload.
• All code runs error-free on my hardware with clear README instructions.

If you’ve built ML-powered SaaS tools or payment-gated content pipelines before, I’d love to see them. Let’s turn draft renders into showroom-ready visuals together." (client-provided description)