Deci## 📦 Contents

• loan-frontend/

  • Modern React + TypeScript frontend with responsive sidebar navigation
  • Redux Toolkit for state management
  • Tailwind CSS for modern UI components
  • Authentication integration with IdentityServer
  • Dashboard with loan statistics and management

• ai-service/

  • Inference service exposing /predict
  • Trains a RandomForest model for approve | pending | reject decisions

• identity-server/

  • Duende IdentityServer + ASP .NET Identity for user management and OAuth2/OIDC

• credit-bureau/

  • (Optional) Client microservice querying a "Credit Bureau API"

• loan-service/

  • ASP .NET Core service for loan request management
  • Listens to UserRegisteredEvent via RabbitMQ and persists Customer
  • Exposes secured endpoints under /loans

• k8s/

  • Manifests for Secrets, ConfigMaps, Deployments, Services, Ingress, StatefulSet (SQL Server), Jobs, CRD Queues, etc.ern microservices suite for credit evaluation and loan management, featuring a React frontend with modern sidebar navigation, FastAPI ML service, ASP .NET Core backend, Duende IdentityServer authentication, and Kubernetes deployment.

---

📦 Contents

• ai-service/

  • Inference service exposing /predict
  • Trains a RandomForest model for approve | pending | reject decisions

• identity-server/

  • Duende IdentityServer + ASP .NET Identity for user management and OAuth2/OIDC

• credit-bureau/

  • (Optional) Client microservice querying a “Credit Bureau API”

• loan-service/

  • ASP .NET Core service for loan request management
  • Listens to UserRegisteredEvent via RabbitMQ and persists Customer
  • Exposes secured endpoints under /loans

• k8s/

  • Manifests for Secrets, ConfigMaps, Deployments, Services, Ingress, StatefulSet (SQL Server), Jobs, CRD Queues, etc.

---

🚀 Architecture

This C4 diagram shows the Context and Container views of the Decision Engine ecosystem:

• Users interact through the Web Frontend ( React.js ).
• Loan Service (.NET 7 + MassTransit) handles loan requests and consumes events from RabbitMQ.
• Identity Server (Duende IdentityServer) provides authentication & authorization.
• AI Service (Python + FastAPI) processes audio summaries and risk scoring.
• RabbitMQ acts as the message bus with retry and DLQ support.
• Loan Database (MongoDB/SQL Server) persists customers and loans.
• Prometheus collects metrics from all services.

---

🔄 Latest Changes

• Modern Frontend Implementation

  • Created responsive React + TypeScript frontend with modern sidebar navigation
  • Implemented Redux Toolkit for state management with loan and auth slices
  • Integrated Tailwind CSS for modern, professional UI components
  • Built modular dashboard with statistics grid, loan requests table, and creation modal
  • Added JWT authentication integration with IdentityServer
  • Configured proper routing and protected routes

• Retry & Dead-Letter Queue

  • Configured MassTransit to use exponential retry (3 attempts, 1s–10s) for UserRegisteredConsumer.
  • Bound failed messages to user-registered-queue-dlq dead-letter queue in RabbitMQ.
  • Updated Kubernetes manifests to declare user-registered-queue and user-registered-queue-dlq CRD queues.

---

⚙️ Prerequisites

• Docker
• kubectl
• Minikube / GKE / AKS / EKS
• .NET 9 SDK (for IdentityServer & Loan-Service)
• Python 3.10+ (for ai-service)

---

🛠️ Local Setup

1. Clone the repo

   git clone https://github.com/your-user/decision-engine.git
   cd decision-engine

2. Run SQL Server locally

   docker run -e "ACCEPT_EULA=Y" -e "SA_PASSWORD=YourStrong@Passw0rd" \
     -p 1433:1433 --name mssql-local -d mcr.microsoft.com/mssql/server:2019-latest

3. Build & run AI-Service

   cd ai-service
   pip install -r requirements.txt
   python training.py         # train model
   uvicorn app:app --reload

4. Run IdentityServer

   cd identity-server
   dotnet run

5. Run Loan-Service

   cd loan-service/src/LoanService.Api
   dotnet run

6. Run Frontend

   cd loan-frontend
   npm install
   npm run dev

---

📦 Docker & Kubernetes

1. Build Docker images

   # AI-Service
   docker build -t yourrepo/ai-inference:latest ai-service/
   
   # IdentityServer
   docker build -t yourrepo/identity-server:latest identity-server/
   
   # Loan-Service
   docker build -t yourrepo/loan-service:latest loan-service/src/

2. Push images to registry

   docker push yourrepo/ai-inference:latest
   docker push yourrepo/identity-server:latest
   docker push yourrepo/loan-service:latest

3. Deploy to Kubernetes

   # Ensure namespace
   kubectl create ns decision-engine-dev || true
   kubectl config set-context --current --namespace=decision-engine-dev
   
   # Apply all k8s manifests
   kubectl apply -f k8s/

4. Verify deployments

   kubectl get all
   kubectl logs deployment/identity-server
   kubectl logs deployment/loan-service
   kubectl port-forward svc/ai-inference 8000:8000

---

🔍 Testing

• AI-Service Predict

  curl http://localhost:8000/predict \
    -H "Content-Type: application/json" \
    -d '{"salary":5000000,"age":30,"credit_score":700,"total_debt":1000000,"payment_history":[{"month":"2025-06","status":"on_time"}]}'

• IdentityServer Metadata Visit http://identity-server.local/.well-known/openid-configuration

• Loan-Service Endpoints

  # list all loans
  curl -H "Authorization: Bearer <token>" http://loan-service.local/loans
  
  # get by id
  curl -H "Authorization: Bearer <token>" http://loan-service.local/loans/<loanId>

---

🎯 Best Practices

• Store secrets in Kubernetes Secrets
• Use Helm for parameterized templating
• Integrate CI/CD pipelines
• Add unit & integration tests
• Monitor with Prometheus & Grafana

---

🛣 Roadmap

• Enhance Loan-Service with more business rules
• ✅ Modern Frontend implemented (React + Tailwind + Redux Toolkit)
• Add RabbitMQ orchestration for multi-step workflows
• Implement advanced dashboard analytics and reporting
• Add unit & integration tests for frontend components
• Improve security, performance and observability

---

⚖️ License

MIT © Cristian Mendoza

🔗 LinkedIn