overview.
An AI-powered SaaS platform where users generate stunning images from text prompts, with a credit-based monetisation system and Razorpay payment integration.
technical implementation.
Atomic Credit Deduction System
Engineered strict database transactions in MongoDB to ensure credits are only deducted from the user's account upon successful verification of the Gemini API image response.
Indian Payment Gateway Webhooks
Implemented secure Razorpay webhook signature verification to handle asynchronous payment success responses and update user credits reliably.
Token-Based Session Management
Built custom Express middleware for stateless JWT verification to protect sensitive image generation and transaction history routes.
Optimized Asset Delivery
Configured direct database image caching and retrieval to speed up gallery load times for user histories.
key features.
- Text-to-Image Generation: High-fidelity image output powered by Google Gemini API integration
- Credit-Based SaaS Model: Complete system with a free registration tier and paid credit top-ups
- Razorpay Gateway: Secure Indian payment gateway integration for seamless transaction processing
- JWT-Based Security: Secure user sessions and role-protected backend endpoints using Bcrypt hashing
- Creation History Gallery: Persistent database storage for browsing, sharing, and re-downloading past creations
- UX Loading Skeleton: Smooth, layout-stable loading states during AI image rendering and delivery
screenshots.
Intuitive prompting interface with dynamic loading skeletons during image generation
Clean workspace workspace highlighting successful credit-based generations
Step-by-step guidance on prompt generation, credit usage, and downloading assets
challenges & solutions.
Challenge: Preventing credit loss on failed API generations
Solution: Wrapped the AI image generation call and the database credit deduction in an idempotent transaction block, restoring credits if an API timeout occurs.
Challenge: Verifying asynchronous payment webhook authenticity
Solution: Developed a robust verification layer utilizing node crypto to validate the Razorpay signature against the raw payload before updating user balances.
Challenge: Maintaining UI layout stability during slow AI processing
Solution: Designed visual skeleton loaders that preserve container dimensions, eliminating cumulative layout shifts while waiting for the image stream.
Challenge: Handling high concurrency during peak image generation
Solution: Established rate-limiting middleware and request throttling on Express servers to distribute traffic and prevent API threshold failures.