In collaboration with Wildlife.LA, an innovative Los Angeles-based company specializing in cutting-edge application experiences, I developed an AI portrait transformation application. This ambitious project allows users to transform their photos into stylized artworks, drawing inspiration from different universes: movies, TV series, animations, or sports contexts.
Project Overview
The AWS Portrait Gallery application represents a significant innovation in AI image generation. At various conventions and events in the United States, the experience has generated tremendous enthusiasm!
Architecture and Technologies
Cloud & Infrastructure
I designed and implemented a cloud-native architecture on AWS, including:
- Optimized image storage via AWS S3
- Process management with DynamoDB
- AI processing with AWS Bedrock (Claude 3.5)
- Advanced integration of ComfyUI for image generation
Frontend Development
The user interface, developed with Next.js, offers a smooth and intuitive experience:
- Instant photo upload
- Interactive selection of artistic styles
- Real-time visualization of the generation process
- Responsive adaptation for all devices
DevOps Pipeline
Implementation of a modern and automated infrastructure:
- CI/CD via GitHub Actions
- Infrastructure as Code with Terraform
- Deployment and monitoring on Vercel
Technical Innovation
Advanced AI System
- Development of a dynamic prompt system
- Sophisticated integration of ComfyUI
- Optimized pipeline for fast, high-quality generations
Performance & Scalability
- Highly available architecture
- Efficient load spike management
- Intelligent queue system
Impact & Results
- Successful presentation at CES 2024 with high attendance
- Development of two major system variations
- Efficient management of high user traffic
- Exceptional feedback from users and professionals
Technical Stack
Frontend Backend Cloud DevOps
- Next.js - API Routes - AWS S3 - GitHub Actions
- React - Node.js - DynamoDB - Terraform
- TailwindCSS - Bedrock - Vercel
- ComfyUI
