Northwind B2B Ordering Portal

A production-quality B2B ordering portal with AI-powered conversational commerce, RAG-based product discovery, and modern web interface. Built with ASP.NET Core 9, Entity Framework Core, and integrated with OpenAI and Qdrant for intelligent product search and natural language ordering.

Prerequisites

.NET 9 SDK
SQL Server LocalDB or SQL Server Express/Full
Optional: SQLite

Solution Structure

Northwind.Portal.sln
├── src/
│   ├── Northwind.Portal.Web/          (ASP.NET Core MVC)
│   ├── Northwind.Portal.Domain/        (Entities, DTOs, Interfaces)
│   ├── Northwind.Portal.Data/          (EF Core, Repositories, Services)
│   └── Northwind.Portal.AI/            (AI Services, Chat, RAG)

Quick Start

Set connection string (User Secrets):

cd src/Northwind.Portal.Web
dotnet user-secrets set "ConnectionStrings:NorthwindsDb" "Server=(localdb)\mssqllocaldb;Database=NorthwindsDb;Trusted_Connection=True;TrustServerCertificate=True;"

Run the application:

dotnet run --project src/Northwind.Portal.Web/Northwind.Portal.Web.csproj

The application will automatically create the database, run migrations, and seed initial data.

Demo Credentials

SuperAdmin: admin@northwind.com / Admin123!
AdminOps: ops@northwind.com / Ops123!
Customer1: customer1@example.com / Customer123! (ALFKI)
Customer2: customer2@example.com / Customer123! (ANATR)

✨ Key Features

🤖 AI-Powered Conversational Commerce

Natural Language Ordering: Chat with AI to search products, add to cart, and place orders
Order History Queries: "Show me my last 3 orders"
Smart Reordering: "Reorder what I bought last month"
Cart Modifications: "Change quantity of product X in my cart"
Shipping Estimates: "When will my order arrive?"
Multi-product Bundles: "Add everything from my last order"
Smart Cart Suggestions: "You might also need Y with X"
RAG-Based Semantic Search: Vector embeddings for intelligent product discovery

🛍️ Customer Portal

Product Catalog: Browse with advanced search, filtering, and pagination
Shopping Cart: Real-time cart management with item count badge
Order Management: View order history, track shipments, and reorder
AI Chat Assistant: Conversational interface for all shopping needs

👨‍💼 Admin Portal

Product Management:
- Create, edit, and view products
- Bulk import from CSV with progress tracking
- Automatic RAG indexing during import
- Reindex all products to vector database
- Advanced search and filtering
Order Management: Full order lifecycle management and fulfillment
Customer Management: Customer accounts and relationship management
Reports & Analytics: Business intelligence and reporting tools

🔍 Vector Search & RAG

Semantic Product Search: Find products using natural language queries
Automatic Embedding Generation: Products indexed with OpenAI embeddings
Qdrant Integration: High-performance vector database for similarity search
Bulk Reindexing: Reindex entire product catalog with progress tracking

AI Configuration

OpenAI Setup

dotnet user-secrets set "LLMOptions:OpenAI:ApiKey" "your-api-key"

Qdrant Setup

docker run -p 6333:6333 -p 6334:6334 qdrant/qdrant

Default configuration: http://localhost:6333

📦 Product Management

CSV Product Import

Location: Admin Portal → Products → Import from CSV

Features:

Upload CSV files with product data
Real-time progress bar during import
Automatic validation and error reporting
Optional RAG indexing to vector store
Download sample CSV template

Required Column: ProductName

Optional Columns: Description, CategoryId, SupplierId, QuantityPerUnit, UnitPrice, UnitsInStock, UnitsOnOrder, ReorderLevel, Discontinued

Sample Format:

ProductName,Description,CategoryId,SupplierId,UnitPrice,UnitsInStock,Discontinued
Organic Green Tea,"Premium organic green tea",1,1,15.50,100,false

Manual Product Management

Create Products: Full product form with all fields
Edit Products: Update existing products with validation
Product Details: View comprehensive product information
Search & Filter: Find products by name, category, supplier, or status

Vector Database Reindexing

Location: Admin Portal → Products → Reindex to Vector DB

Reindex all products to vector database for RAG search
Real-time progress tracking with statistics
Error reporting and detailed results
Automatic embedding regeneration

Roles

CustomerUser: Customer portal access
CustomerApprover: Customer portal with approvals
AdminOps: Order and customer management
AdminCatalog: Product management
AdminFulfillment: Order fulfillment
SuperAdmin: Full system access

Order Status Flow

Submitted → PendingApproval → Approved → Picking → Shipped → Cancelled/Rejected

Configuration

Configuration priority:

User Secrets (Development) / Environment Variables (Production)
appsettings.{Environment}.json
appsettings.json

Connection strings must be set via User Secrets or environment variables.

🛠️ Technology Stack

Backend: ASP.NET Core 9, Entity Framework Core
AI/ML: OpenAI GPT-4, Bipins.AI library (v1.0.4)
Vector Database: Qdrant
Frontend: Bootstrap 5, Font Awesome 6, jQuery
Database: SQL Server / LocalDB
Logging: Serilog with structured logging

🔧 Troubleshooting

Migrations: Run manually if needed:

dotnet ef database update --project src/Northwind.Portal.Data/Northwind.Portal.Data.csproj --context NorthwindDbContext

CSV Import:

Ensure CategoryId/SupplierId exist in database
Check error messages on import page
Verify CSV format matches sample template

RAG Search:

Verify Qdrant is running: docker ps | grep qdrant
Check OpenAI API key is configured
System falls back to text search if vector search fails
Use "Reindex to Vector DB" to rebuild embeddings

Product Name Length: ProductName supports up to 100 characters (migration applied automatically)

Docker

Local Development

docker-compose up -d

Exposes app on http://localhost:8080

Production Image

The application is automatically built and pushed to GitHub Container Registry via GitHub Actions.

Pull the latest image:

docker pull ghcr.io/bipindra/northwinds.ai:latest

Run the container:

docker run -d -p 8080:8080 \
  -e ConnectionStrings__NorthwindsDb="your-connection-string" \
  -e ASPNETCORE_ENVIRONMENT=Production \
  ghcr.io/bipindra/northwinds.ai:latest

Available tags:

latest - Latest build from master branch
master-<sha> - Specific commit
v1.0.0 - Semantic version tags

Image registry: ghcr.io/bipindra/northwinds.ai

☸️ Kubernetes (Local Deployment)

Deploy the Northwind Portal to a local Kubernetes cluster. The manifests are split into three layers — base, infrastructure, and app — so you can deploy the full stack locally or deploy only the app against existing SQL Server and Qdrant instances.

Prerequisites

Docker Desktop with Kubernetes enabled, or minikube
kubectl CLI installed and configured
Docker CLI for building images

Directory Layout

k8s/
├── base/                            # Shared cluster resources
│   └── namespace.yaml               # Kubernetes namespace
├── infrastructure/                   # Deployable independently; skip when using external services
│   ├── secrets.yaml                 # SQL Server credentials
│   ├── sqlserver.yaml               # SQL Server Deployment + Service + PVC
│   └── qdrant.yaml                  # Qdrant Deployment + Service + PVC
├── app/                              # Deployable independently against any SQL Server + Qdrant
│   ├── secrets.yaml                 # Application connection string and API keys
│   ├── configmap.yaml               # Application environment configuration
│   └── deployment.yaml             # Application Deployment + NodePort Service
├── deploy.ps1                        # One-click deploy (PowerShell)
├── deploy.sh                         # One-click deploy (Bash)
├── teardown.ps1                      # Teardown (PowerShell)
└── teardown.sh                       # Teardown (Bash)

Manifest Reference

1. `k8s/base/namespace.yaml` — Namespace

Creates the northwind namespace that all other resources belong to. This must be applied first.

Field	Value
Kind	`Namespace`
Name	`northwind`

kubectl apply -f k8s/base/namespace.yaml

2. `k8s/infrastructure/secrets.yaml` — Infrastructure Secrets

Stores credentials used by the SQL Server container. Only needed when deploying infrastructure locally.

Key	Description	Default
`SA_PASSWORD`	SQL Server SA password	`YourStrong@Passw0rd`
`ConnectionStrings__NorthwindsDb`	Connection string using the in-cluster `sqlserver` service	Points to `Server=sqlserver`
`LLMOptions__OpenAI__ApiKey`	(commented out) OpenAI API key for AI features	—

Customize before applying:

Change SA_PASSWORD and update the matching password in ConnectionStrings__NorthwindsDb.
Optionally uncomment and set LLMOptions__OpenAI__ApiKey.

kubectl apply -f k8s/infrastructure/secrets.yaml

3. `k8s/infrastructure/sqlserver.yaml` — SQL Server

Deploys SQL Server 2022 Developer Edition with persistent storage. Contains three Kubernetes resources:

Resource	Kind	Name	Details
Database server	`Deployment`	`sqlserver`	1 replica, image `mcr.microsoft.com/mssql/server:2022-latest`, port `1433`
Internal endpoint	`Service` (ClusterIP)	`sqlserver`	Accessible at `sqlserver:1433` within the cluster
Data volume	`PersistentVolumeClaim`	`sqlserver-pvc`	2 Gi, `ReadWriteOnce`, mounted at `/var/opt/mssql`

Resource limits: 250m–1000m CPU, 512Mi–2Gi memory.

Depends on: k8s/infrastructure/secrets.yaml (reads SA_PASSWORD via secretKeyRef).

kubectl apply -f k8s/infrastructure/sqlserver.yaml

Verify:

kubectl -n northwind wait --for=condition=ready pod -l app=sqlserver --timeout=120s
kubectl -n northwind logs -f deployment/sqlserver

4. `k8s/infrastructure/qdrant.yaml` — Qdrant Vector Database

Deploys the Qdrant vector database for RAG-based semantic product search. Contains three Kubernetes resources:

Resource	Kind	Name	Details
Vector DB	`Deployment`	`qdrant`	1 replica, image `qdrant/qdrant:latest`, ports `6333` (HTTP) and `6334` (gRPC)
Internal endpoint	`Service` (ClusterIP)	`qdrant`	Accessible at `qdrant:6333` / `qdrant:6334` within the cluster
Data volume	`PersistentVolumeClaim`	`qdrant-pvc`	1 Gi, `ReadWriteOnce`, mounted at `/qdrant/storage`

Resource limits: 100m–500m CPU, 256Mi–1Gi memory.

No dependencies — this manifest is self-contained.

kubectl apply -f k8s/infrastructure/qdrant.yaml

Verify:

kubectl -n northwind wait --for=condition=ready pod -l app=qdrant --timeout=60s
kubectl -n northwind logs -f deployment/qdrant

5. `k8s/app/secrets.yaml` — Application Secrets

Stores sensitive application configuration injected as environment variables into the app pod.

Key	Description	Default
`ConnectionStrings__NorthwindsDb`	SQL Server connection string	Points to in-cluster `sqlserver` service
`LLMOptions__OpenAI__ApiKey`	(commented out) OpenAI API key	—

Customize before applying:

Using local infrastructure: The default connection string (Server=sqlserver;...) works as-is.
Using external SQL Server: Replace the connection string with your external server's host, credentials, and database name.
AI features: Uncomment LLMOptions__OpenAI__ApiKey and set your key.

kubectl apply -f k8s/app/secrets.yaml

6. `k8s/app/configmap.yaml` — Application Configuration

Non-sensitive application settings injected as environment variables into the app pod.

Key	Description	Default
`ASPNETCORE_ENVIRONMENT`	ASP.NET Core environment	`Development`
`ASPNETCORE_URLS`	Kestrel listen URL	`http://+:8080`
`DatabaseOptions__DbProvider`	Database provider	`SqlServer`
`VectorDbOptions__Provider`	Vector DB provider	`Qdrant`
`VectorDbOptions__Qdrant__Endpoint`	Qdrant HTTP endpoint	`http://qdrant:6333` (in-cluster)
`VectorDbOptions__Qdrant__CollectionName`	Qdrant collection name	`northwind_portal`
`VectorDbOptions__Qdrant__CreateCollectionIfMissing`	Auto-create collection	`true`

Customize before applying:

Using local infrastructure: Defaults work as-is.
Using external Qdrant: Update VectorDbOptions__Qdrant__Endpoint to your external host (e.g. http://your-qdrant-host:6333).
Production: Change ASPNETCORE_ENVIRONMENT to Production.

kubectl apply -f k8s/app/configmap.yaml

7. `k8s/app/deployment.yaml` — Application Deployment & Service

Deploys the Northwind Portal web application. Contains two Kubernetes resources:

Resource	Kind	Name	Details
Web app	`Deployment`	`northwind-portal`	1 replica, image `northwind-portal:latest`, port `8080`
External access	`Service` (NodePort)	`northwind-portal`	Maps port `8080` → NodePort `30080`

Deployment details:

Image pull policy: Never — expects a locally built Docker image (not pulled from a registry).
Environment: All variables injected from northwind-config (ConfigMap) and northwind-app-secrets (Secret) via envFrom.
Readiness probe: GET /health on port 8080, starts after 15s, checks every 10s.
Liveness probe: GET /health on port 8080, starts after 30s, checks every 15s.
Resource limits: 100m–500m CPU, 256Mi–512Mi memory.

Depends on:

k8s/app/configmap.yaml (referenced as northwind-config)
k8s/app/secrets.yaml (referenced as northwind-app-secrets)
A locally built Docker image named northwind-portal:latest

Build the image first:

docker build -t northwind-portal:latest .

# If using minikube, also load the image:
minikube image load northwind-portal:latest

Apply:

kubectl apply -f k8s/app/deployment.yaml

Verify:

kubectl -n northwind wait --for=condition=ready pod -l app=northwind-portal --timeout=180s
kubectl -n northwind logs -f deployment/northwind-portal

Deployment Scenarios

Full Stack (all manifests)

Deploy everything — SQL Server, Qdrant, and the app — to a clean local cluster.

Apply order:

kubectl apply -f k8s/base/namespace.yaml
kubectl apply -f k8s/infrastructure/secrets.yaml
kubectl apply -f k8s/infrastructure/sqlserver.yaml
kubectl apply -f k8s/infrastructure/qdrant.yaml
kubectl apply -f k8s/app/secrets.yaml
kubectl apply -f k8s/app/configmap.yaml
kubectl apply -f k8s/app/deployment.yaml

Or use the deploy script:

# PowerShell
.\k8s\deploy.ps1

# Bash
chmod +x k8s/deploy.sh
./k8s/deploy.sh

The script automatically builds the Docker image, loads it into minikube (if detected), applies all manifests in order, and waits for pods to become ready.

App Only (existing SQL Server & Qdrant)

When SQL Server and Qdrant already exist (managed services, shared cluster, etc.), skip the infrastructure manifests entirely.

Edit k8s/app/secrets.yaml — set ConnectionStrings__NorthwindsDb to your external SQL Server.
Edit k8s/app/configmap.yaml — set VectorDbOptions__Qdrant__Endpoint to your external Qdrant.
Apply:

kubectl apply -f k8s/base/namespace.yaml
kubectl apply -f k8s/app/secrets.yaml
kubectl apply -f k8s/app/configmap.yaml
kubectl apply -f k8s/app/deployment.yaml

Or use the deploy script:

.\k8s\deploy.ps1 -AppOnly

./k8s/deploy.sh --app-only

Infrastructure Only

Prepare a shared SQL Server + Qdrant environment without deploying the app (e.g. for other teams to use).

kubectl apply -f k8s/base/namespace.yaml
kubectl apply -f k8s/infrastructure/secrets.yaml
kubectl apply -f k8s/infrastructure/sqlserver.yaml
kubectl apply -f k8s/infrastructure/qdrant.yaml

Or use the deploy script:

.\k8s\deploy.ps1 -InfraOnly

./k8s/deploy.sh --infra-only

Access the Application

Method	URL
Docker Desktop Kubernetes	http://localhost:30080
minikube	Run `minikube service northwind-portal -n northwind`

Useful Commands

# View app logs
kubectl -n northwind logs -f deployment/northwind-portal

# View SQL Server logs
kubectl -n northwind logs -f deployment/sqlserver

# View Qdrant logs
kubectl -n northwind logs -f deployment/qdrant

# Open a shell in the app container
kubectl -n northwind exec -it deployment/northwind-portal -- /bin/bash

# Scale the app (infrastructure should remain at 1 replica)
kubectl -n northwind scale deployment northwind-portal --replicas=3

# Restart the app after config/secret changes
kubectl -n northwind rollout restart deployment northwind-portal

# Check pod events (useful for debugging startup issues)
kubectl -n northwind describe pod -l app=northwind-portal

# View all resources in the namespace
kubectl -n northwind get all

Teardown

PowerShell:

.\k8s\teardown.ps1              # Remove everything
.\k8s\teardown.ps1 -AppOnly     # Remove only the app, keep infrastructure
.\k8s\teardown.ps1 -InfraOnly   # Remove only infrastructure, keep app

Bash:

./k8s/teardown.sh               # Remove everything
./k8s/teardown.sh --app-only    # Remove only the app, keep infrastructure
./k8s/teardown.sh --infra-only  # Remove only infrastructure, keep app

Manual (remove everything at once):

kubectl delete namespace northwind

Note: Deleting the namespace removes all resources including persistent volume claims. Database and vector data will be lost.

License

Demonstration project based on the Northwind database schema.

Name		Name	Last commit message	Last commit date
Latest commit History 19 Commits
.github/workflows		.github/workflows
k8s		k8s
src		src
.dockerignore		.dockerignore
.gitignore		.gitignore
Dockerfile		Dockerfile
LICENSE.txt		LICENSE.txt
Northwind.Portal.sln		Northwind.Portal.sln
README.md		README.md
apply-migration.ps1		apply-migration.ps1
azure-pipelines-simple.yml		azure-pipelines-simple.yml
azure-pipelines.yml		azure-pipelines.yml
docker-compose.yml		docker-compose.yml
startup_logs.txt		startup_logs.txt

Folders and files

Latest commit

History

Repository files navigation

Northwind B2B Ordering Portal

Prerequisites

Solution Structure

Quick Start

Demo Credentials

✨ Key Features

🤖 AI-Powered Conversational Commerce

🛍️ Customer Portal

👨‍💼 Admin Portal

🔍 Vector Search & RAG

AI Configuration

OpenAI Setup

Qdrant Setup

📦 Product Management

CSV Product Import

Manual Product Management

Vector Database Reindexing

Roles

Order Status Flow

Configuration

🛠️ Technology Stack

🔧 Troubleshooting

Docker

Local Development

Production Image

☸️ Kubernetes (Local Deployment)

Prerequisites

Directory Layout

Manifest Reference

1. k8s/base/namespace.yaml — Namespace

2. k8s/infrastructure/secrets.yaml — Infrastructure Secrets

3. k8s/infrastructure/sqlserver.yaml — SQL Server

4. k8s/infrastructure/qdrant.yaml — Qdrant Vector Database

5. k8s/app/secrets.yaml — Application Secrets

6. k8s/app/configmap.yaml — Application Configuration

7. k8s/app/deployment.yaml — Application Deployment & Service

Deployment Scenarios

Full Stack (all manifests)

App Only (existing SQL Server & Qdrant)

Infrastructure Only

Access the Application

Useful Commands

Teardown

License

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Uh oh!

Contributors

Uh oh!

Languages

1. `k8s/base/namespace.yaml` — Namespace

2. `k8s/infrastructure/secrets.yaml` — Infrastructure Secrets

3. `k8s/infrastructure/sqlserver.yaml` — SQL Server

4. `k8s/infrastructure/qdrant.yaml` — Qdrant Vector Database

5. `k8s/app/secrets.yaml` — Application Secrets

6. `k8s/app/configmap.yaml` — Application Configuration

7. `k8s/app/deployment.yaml` — Application Deployment & Service

Packages