AIMAAK is an AI-powered conversational assistant designed for real-time multilingual customer support (Darija/French). It leverages FastAPI, LangChain, Redis, and vector databases to deliver dynamic, context-aware, and scalable responses. The backend exposes RESTful APIs for chat interactions, document-based Q&A, and relational database integrations.

---

🚀 Features

• AI-Powered Responses: Uses LangChain with Google Generative AI for smart and contextual replies.
• Multilingual Support: Understands Moroccan Darija and French.
• Real-Time Chat: Backend optimized with FastAPI and Redis for fast response handling.
• Document Q&A: Extracts knowledge from PDF documents (e.g., Lara_Fashion_QA.pdf) using vector stores and embeddings.
• Relational DB Integration: Provides SQL-based knowledge querying.
• Cross-Origin Ready: CORS enabled for easy frontend integration.
• Voice & Chat Modes: Supports both chatbot widget mode and voice-based interaction using Gemini API.

---

🛠 Tech Stack

• Framework: FastAPI
• AI/LLM Framework: LangChain + OpenAI & Gemini
• Database: MySQL + Redis (for session and cache handling)
• Vector Store: FAISS
• Deployment: Uvicorn workers

---

📦 Installation

Prerequisites

• Python 3.10+
• SQL Engine
• Redis
• LangChain
• FAISS
• FastAPI
• virtualenv (recommended)

Steps

# 1. Clone the repository
git clone https://github.com/<your-username>/AIMaak-Backend.git
cd AIMaak-Backend

# 2. Create a virtual environment
python -m venv venv
source venv/bin/activate  # Windows: venv\\Scripts\\activate

# 3. Install dependencies
pip install -r requirements.txt

# 4. Configure environment variables
cp .env.example .env
# Edit .env and add your REDIS, MySQL, and API keys

# 5. Start the FastAPI server
uvicorn main:app --reload

⚙️ Environment Variables

Create a .env file in the project root with:

REDIS_HOST=localhost
REDIS_PORT=6379
REDIS_PASSWORD=your_password

DB_URI=your_db_url_if_you_are_using_an_oline_db

MYSQL_HOST=localhost
MYSQL_USER=root
MYSQL_PASSWORD=your_password
MYSQL_DB=aimaak

GEN_API_KEY=your_key
OPENAI_API_KEY=sk-your_key

🔗 API Endpoints

Method	Endpoint	Description
POST	/chat	Send a chat message to AI
POST	/documents/query	Ask a question on uploaded docs
POST	/sql/query	Converts a natural language question to a safe SQL query, runs it, and returns the results with a natural language answer

📜 License

This project is licensed under the MIT License.

👥 Contributors

• Zakaria EL-AOUFI: Created all RAG pipelines
• Wassim Midi: Deployed and integrated voice into AIMAAK
• Mohammed Amine El Abiad: Frontend Developer