More3zdenAI

More3zdenAI is a RAG-powered portfolio assistant that lets visitors ask natural-language questions about a person's skills, projects, and experience. The app combines a FastAPI backend, a Next.js frontend, FAISS-based retrieval, and a local Ollama model to deliver grounded answers with source attribution.

It is designed as a modern, containerized AI product with streaming responses, persistent conversations, and optional voice output.

---

Overview

More3zdenAI replaces static portfolio navigation with an interactive chat experience. Users can ask questions such as:

• What technologies do you work with?
• What projects have you built?
• What experience do you have in backend development?

The backend retrieves relevant chunks from a markdown knowledge base, generates an answer with a local LLM, and returns the results to the frontend with source references.

---

Architecture

User → Next.js UI → FastAPI API → RAG Pipeline
                              ├─ FAISS vector search
                              ├─ PostgreSQL conversation history
                              ├─ Redis response cache
                              └─ Ollama LLM generation

Core components

• Frontend: Next.js 14, React, TypeScript, Tailwind CSS
• Backend: Python, FastAPI, SQLAlchemy, Pydantic
• AI / RAG: FAISS, sentence-transformers, Ollama, local embeddings
• Data: PostgreSQL for conversations, Redis for caching
• Infra: Docker Compose, Nginx, optional TTS service

---

Features

• Conversational RAG over a portfolio knowledge base
• Grounded answers with source attribution
• Streaming chat responses using Server-Sent Events
• Persistent chat sessions in PostgreSQL
• Redis-based caching for repeat queries
• Optional text-to-speech using LuxTTS with a local fallback
• Health endpoint for service monitoring
• Docker-based deployment for local and production-friendly setups

---

Project Structure

.
├── backend/
│   ├── app/
│   │   ├── main.py
│   │   ├── config.py
│   │   ├── database.py
│   │   ├── models.py
│   │   ├── schemas.py
│   │   ├── cache.py
│   │   ├── rag/
│   │   └── routes/
│   ├── requirements.txt
│   └── Dockerfile
├── frontend/
│   ├── src/
│   │   ├── app/
│   │   ├── components/
│   │   └── lib/
│   ├── package.json
│   └── Dockerfile
├── knowledge_base/
│   └── portfolio.md
├── tts/
├── docker/
├── scripts/
├── docker-compose.yml
└── .env.example

---

Getting Started

Prerequisites

• Docker Desktop or Docker Engine with Docker Compose
• Git
• At least 4 GB of free disk space for the model and dependencies

1. Clone the repository

git clone https://github.com/3zden/More3zdenAi.git
cd More3zdenAi

2. Configure environment

cp .env.example .env

Adjust the values in .env if needed. The defaults are suitable for local development.

3. Start the stack

./scripts/start.sh

Or run the services directly:

docker compose up -d --build

4. Open the app

• UI: http://localhost
• API docs: http://localhost:8000/docs
• Health check: http://localhost/api/health/

---

Running the Services

Backend logs

docker compose logs -f backend

Frontend logs

docker compose logs -f frontend

Stop everything

docker compose down

Full reset

docker compose down -v

---

Updating the Knowledge Base

Edit knowledge_base/portfolio.md, then rebuild the FAISS index and restart the backend:

docker compose exec backend python -c "import os; os.remove('/app/data/faiss.index'); os.remove('/app/data/chunks.pkl')"
docker compose restart backend

The backend will rebuild the index on startup if the cached files are missing.

---

API Endpoints

Chat

• POST /api/chat/
  • Non-streaming chat request with persisted conversation history and caching
• GET /api/chat/stream/
  • Server-Sent Events endpoint for streaming responses

Conversation

• GET /api/conversation/{session_id}/
  • Retrieve the full conversation history for a session

Health

• GET /api/health/
  • Returns backend and service health status

TTS

• POST /api/tts/
  • Synthesizes speech from text using the optional LuxTTS service

---

Environment Variables

The app uses environment variables from .env. Key variables include:

---

Development Notes

• The backend expects its supporting services to be reachable via the Docker network hostnames such as db, redis, and ollama.
• The FAISS index is built on first startup and persisted in the faiss_data volume.
• The TTS service is optional and can be enabled with the voice profile in Docker Compose.

---

License

This project is intended for personal portfolio and demonstration use.