Overview
AI-powered professional headshot generator serving users across Europe and North America. The platform transforms casual selfies into professional portraits for LinkedIn, CV, and social media in under 30 seconds. Architected a microservices backend with 6 independent services using Node.js/Fastify, with PostgreSQL for persistent storage and RabbitMQ for asynchronous job queuing. Built an ML pipeline proxy via LiteLLM to orchestrate requests across multiple image generation models, with face detection preprocessing for consistent output quality. Implemented real-time progress streaming via Server-Sent Events, eliminating polling overhead. Designed a credit-based payment system with Stripe integration supporting three tiered packages (Starter, Professional, Executive). Deployed on Kubernetes with Helm charts, CI/CD via GitHub Actions, and enterprise-grade secrets management. Monitoring stack includes VictoriaMetrics and Grafana dashboards with structured logging via VictoriaLogs. Supports 10 languages with GDPR-compliant 30-day automatic data retention and end-to-end encryption.
Technology Stack
Key Features
- •AI-powered headshot generation with multiple professional styles
- •Real-time progress updates via Server-Sent Events (SSE)
- •Credit-based payment system with Stripe integration
- •Multi-language support (10 languages: EN, RU, ES, DE, FR, IT, PT, ZH, JA, KO)
- •24-hour result access with automatic cleanup
- •Social sharing capabilities
- •Email notifications for job completion
- •Admin dashboard for ML model evaluation
Technical Highlights
- •Microservices architecture with 6 independent services
- •Event-driven architecture using RabbitMQ message queues
- •Kubernetes deployment with Helm charts and GitOps
- •CI/CD pipeline with GitHub Actions
- •Monitoring with VictoriaMetrics and Grafana dashboards
- •SOPS-encrypted secrets management
- •Horizontal scaling with Kubernetes HPA
- •Structured logging with VictoriaLogs
- •Comprehensive test coverage (unit, integration, e2e)
Screenshots
