AI-Powered Microservice Architecture

🛒 AI-Powered E-Shop Microservices Architecture with .NET Aspire & GenAI

This project demonstrates the development of a cloud-native, AI-powered E-Commerce platform using the .NET Aspire framework, integrating Generative AI capabilities with Microsoft.Extensions.AI, Semantic Kernel, and Ollama's Llama/Phi models.

The system is composed of modular microservices such as Catalog and Basket, powered by PostgreSQL, Redis, and RabbitMQ, and secured with Keycloak. It also features intelligent product discovery through semantic search and Q&A chatbots, enabled by RAG (Retrieval-Augmented Generation) flows.

---

🚀 What You'll Learn

• Build cloud-native distributed architectures with .NET Aspire
• Implement microservices using PostgreSQL, Redis, RabbitMQ
• Architect event-driven communication using MassTransit and RabbitMQ
• Secure endpoints using Keycloak with JWT Bearer Authentication
• Containerize and deploy services to Azure Container Apps
• Power intelligent features with GenAI, Semantic Kernel, and VectorDB
• Integrate Q&A Chatbot and Semantic Product Search using Ollama

---

🧩 Architecture Overview

• Catalog Microservice

  • Stores product data in PostgreSQL
  • Publishes events like ProductPriceChanged to RabbitMQ

• Basket Microservice

  • Uses Redis to manage session-based shopping cart data
  • Subscribes to Catalog events via RabbitMQ to sync basket prices
  • Secured using Keycloak + JWT

• Client Application

  • Built using Blazor WebAssembly
  • Communicates with backend services via REST and messaging

• Messaging

  • Pub/Sub with RabbitMQ
  • Resilient error handling, retries, and routing logic via MassTransit

• Generative AI

  • Embedded Q&A chatbot powered by Semantic Kernel + Ollama
  • RAG-based product semantic search using Vector Store and All-MiniLM embeddings

---

🧠 AI Capabilities

🔍 Semantic Product Search (RAG Flow)

• Generate embeddings for product descriptions using Ollama's All-MiniLM
• Store and query a VectorDB to enable real-time semantic product matching

💬 Q&A Chatbot

• Use Semantic Kernel to create prompt workflows
• Integrate Ollama (Llama/Phi) to answer contextual queries from users

---

🔒 Security

• Integrated Keycloak for identity management
• JWT Bearer tokens used to protect endpoints in the Basket service

---

🐳 Deployment

• All services are containerized using Docker
• Deploy to Azure Container Apps using:

  azd up     # Deploy
  azd down   # Teardown