TL;DR — Text-to-Image Service

Design a text-to-image generation service similar to DALL-E or Midjourney. Users submit text prompts and receive AI-generated images. The system must manage GPU-intensive diffusion model inference, implement content safety filtering on both prompts and generated images, and serve millions of generation requests daily. Key features: Generate images from text prompts with configurable parameters. Support multiple models and aspect ratios.