RAG-Weather-Assistant

This project is a RAG micro-service designed to provide real-time weather information using advanced AI techniques, including Azure OpenAI models,Langchain orchestration and Weaviate vector storage.

📖 Overview

---

✨ Features

• Openai GPT model for generative responses
• Openai embedding model for vector embeddings
• Clean architecture with typed LangChain layers
• 🏗️ Integration with Azure OpenAI model deployment and Weaviate vector database
• 🐍 Flask-based backend with Swagger UI
• Docker and Docker Compose for infrastructure management

---

📁 Project Hierarchy

The project structure follows a clean, modular approach:

├── src/
│   ├── api/                # RESTful API and Swagger documentation
│   ├── adapters/           # Integration layer with external services (e.g., Weaviate)
│   ├── controllers/        # Orchestration layer, managing request handling and response
│   ├── embeddings_providers/ # Language models and embeddings services (Azure-based)
│   ├── llm_providers/      # Language model wrappers for seamless integration
│   ├── services/           # Core business logic, including data indexing and QA systems
│   ├── config/             # Configuration management for runtime settings
│   └── prompts/            # Predefined prompt templates for consistent LLM interactions
├── Dockerfile              # Docker configuration for containerized deployment
├── docker-compose.yml      # Local development environment setup (API + Weaviate)

---

🚀 Quick Start

# 1 – clone
git clone https://github.com/MaximPyanin/rag-weather-assistant.git
cd rag-weather-assistant

# 2 – create .env (fill Azure creds)
cp .env.sample .env
# AZURE_OPENAI_ENDPOINT=https://<resource>.openai.azure.com/
# AZURE_OPENAI_KEY=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
# AZURE_OPENAI_EMBEDDING_ENDPOINT=https://<resource>.openai.azure.com/

# 3 – build & run
docker compose up -d --build

# 4 – open docs
open http://localhost:5000/swagger/

# 5 – sample request
curl -X POST http://localhost:5000/api/v1/ask \
     -H "Content-Type: application/json" \
     -d '{"query": "What is the temperature in Madrid?"}'

---

🔧 Configuration

You can configure the application using environment variables. The following variables are required:

AZURE_OPENAI_ENDPOINT -	Base URL of your Azure OpenAI resource
AZURE_OPENAI_KEY - Secret key or leave empty ↠ Managed Identity token
AZURE_OPENAI_LLM_DEPLOYMENT - GPT deployment (default gpt-35-turbo)
AZURE_OPENAI_EMBEDDING_ENDPOINT - Same resource or separate embeddings endpoint
AZURE_OPENAI_EMBEDDING_DEPLOYMENT - Embedding deployment (default text-embedding-ada-002)

Any value in AppConfig can be overridden via .env

📜 LICENSE

This project is licensed under the MIT License. See the LICENSE file for details.