Enterprise RAG Demo | Bravado Solutions

Building Scalable AI Systems for the Modern Enterprise

Bravado Solutions is an enterprise software development company building scalable AI systems, SaaS platforms, and cloud-native applications.

This repository demonstrates an enterprise-ready Retrieval-Augmented Generation (RAG) architecture on Azure. It includes both C# and Python implementations for integrating proprietary data with GPT models via Azure OpenAI and Azure AI Search.

---

🚀 Overview

Most AI projects fail not because of models—but because of data fragmentation and weak architecture. This reference implementation focuses on:

• Grounding: Ensuring LLM responses are based strictly on enterprise data.
• Scalability: Building RAG pipelines that handle production loads on Azure.
• Hybrid Search: Integrating vector search with GPT models for high accuracy.

---

🧠 Architecture

The solution follows a standard enterprise RAG pipeline:

1. Data Ingestion: Uploading unstructured documents to Azure Storage.
2. Indexing: Azure AI Search handles vectorization and chunking.
3. Retrieval: Context is fetched via vector/hybrid search based on user intent.
4. Generation: Azure OpenAI generates grounded responses using the retrieved context.

---

📂 Repository Structure

enterprise-rag-demo/
├── CSharp/
│   ├── ownData.cs           # C# implementation for data ingestion
│   └── appsettings.json     # Configuration for .NET environment
├── Python/
│   ├── ownData.py           # Python implementation for data ingestion
│   └── python.env           # Environment variables template
├── docs/
│   └── instructions.md      # Detailed setup and usage guides
├── .gitignore               # Excludes secrets and build artifacts
├── LICENSE                  # Repository licensing information
└── README.md                # Project overview and quick start

⚙️ Prerequisites

Azure Subscription You will need an Azure subscription with access to the following resources:

• Azure OpenAI Service: Required for GPT-4o (Chat) and Text-Embedding-3-Small (Embeddings).
• Azure AI Search: To host the vector index and perform retrieval.
• Azure Storage Account: For hosting source documents and training data.

Development Tools

• .NET 7 SDK: For running the C# implementation.
• Python 3.11+: For running the Python implementation and scripts.
• VS Code: (Or your preferred IDE) with the Azure and Python extensions installed.

---

🔧 Setup

1. Provision Azure Resources

• Azure OpenAI: Deploy a chat model (e.g., GPT-4o) and an embedding model.
• Azure AI Search: Create an index with vector support.
• Storage: Upload your PDFs or docs to a container named data.

---

2. Configure Environment

Python:
Run cp .env.example .env and update .env with your Azure endpoints and keys.

C#:
Run Copy-Item appsettings.json.template appsettings.json and update appsettings.json with your Azure configuration.

---

3. Install Dependencies

Python:
Run pip install -r requirements.txt

C#:
Run dotnet add package Azure.AI.OpenAI and dotnet add package Azure.Search.Documents

---

4. Run the Application

Python: Run python ownData.py
C#: Run dotnet run

---

🔐 Security Best Practices

• Secrets: Never commit .env or appsettings.json with real keys.
• Identity: Use Managed Identities (RBAC) instead of API keys for production.
• Rotation: Regularly rotate Azure service keys.

---

🤝 Work with Bravado Solutions

We help enterprises move from AI experimentation to production-grade systems.

• 🌐 Website: bravadosolutions.com
• 📧 Contact: contact@bravadosolutions.com