AIYIKES

Overview

AIYIKES is a complete AI image generation ecosystem. Users can generate images using three different AI models (Flux - Balanced, GPTImage - Premium, Nanobanana - Ultra-fast), choose between public community sharing or private generation, and remix existing images with new prompts. The platform tracks the entire remix lineage, showing how images evolve from original to remixed versions in the dedicated "Ur-Remixes" section.

Problem

Existing AI image generation platforms are either too limiting (usage caps, single models) or lack community features. Users can't build upon each other's work or explore variations of promising images, and most platforms don't preserve the creative lineage of generated content.

Solution

I built a full-stack platform using Next.js and PostgreSQL (Neon) that offers three AI models for different use cases, public/private generation modes, and a unique remix system. When users remix an image, the platform preserves both the original and new prompts, stores the relationship in the database, and displays the complete evolution in a dedicated 'Ur-Remixes' section. Redis caching optimizes gallery performance, and the architecture supports future credit-based monetization.

System Architecture

View Diagram

Core Features

• Three AI models: Flux (Balanced), GPTImage (Premium), Nanobanana (Ultra-fast)
• Public/private image visibility with one-click toggle
• Community gallery with like functionality
• Advanced remix workflow: original image + new prompt = remix
• Ur-Remixes dashboard showing all remixed images with lineage
• User authentication with avatar upload and password management
• Redis caching for user sessions and high-traffic gallery feeds
• Download functionality for all generated and remixed images

Key Engineering Decisions

• Chose Next.js full-stack architecture to unify frontend/backend logic with type-safe API routes
• Implemented Redis for caching user sessions (ID, email) and gallery data, reducing database calls by 80%
• Designed a relational database schema to track remix lineage (original image → prompt → remixed image)
• Used Prisma ORM with PostgreSQL (Neon) for type-safe queries and serverless connection pooling
• Integrated Cloudinary for avatar storage with automatic optimization and transformations
• Built three-model system (Pollinations AI) to balance speed, quality, and cost per generation
• Architecture prepared for credit-based billing by tracking generation costs per user in transaction table

Tech Stack

Demo