AI RAG Showcase

This repository is a showcase project demonstrating Retrieval-Augmented Generation (RAG) using different datasets.
It highlights how RAG can be applied to multiple domains: programming documentation, custom datasets, and even synthetic (fictional) data.

---

Features

• Universal backend in Node.js/TypeScript (Express).
• Qdrant as the vector database for embeddings.
• Worker service for ingestion (document parsing + embeddings).
• Telegram bot built with telegraf.js.
• Next.js web app for testing queries and managing datasets.
• Easily extensible: just drop a new dataset into datasets/ and re-ingest.

---

Project Structure

ai-rag-showcase/
├── apps/
│   ├── api/              # RAG API (Express).
│   ├── tg-bot/           # Telegram bot (telegraf.js).
│   ├── web/              # Next.js panel.
│   └── worker/           # Ingestion worker (BullMQ + Redis).
├── datasets/
│   └── mythlang/         # Fictional "MythLang" programming language.
├── packages/
│   ├── core/             # Shared RAG pipeline logic.
│   ├── embeddings/       # Embeddings providers (OpenAI/local).
│   ├── llm/              # LLM providers (OpenAI/Ollama).
│   ├── utils/            # Config, logger, helpers.
│   └── vectorstore/      # Qdrant wrapper.
├── docker-compose.yml
├── LICENSE
└── README.md

---

Datasets Included

• MythLang → a fictional programming language for spells and rituals (synthetic dataset).

---

Tech Stack

• TypeScript
• MERN → MongoDB, Express.js, React/Next.js, Node.js.
• Qdrant → Vector database for embeddings.
• Redis + BullMQ → Job queue for ingestion.
• OpenAI API → Embeddings + LLM (GPT-4o-mini).
• telegraf.js → Telegram bot.

---

Quick Start

git clone https://github.com/an-ivannikov/ai-rag-showcase
cd ai-rag-showcase

# Install dependencies
pnpm install

# Copy env file
cp .env.example .env

# Start vector DB + Redis + Mongo
docker compose --profile infra  --profile app build --no-cache
docker compose --profile infra  --profile app up --force-recreate -d

# Run backend + worker + bot
pnpm dev

curl -X POST http://localhost:3001/ingest \
  -H "Content-Type: application/json" \
  -d '{"dir":"./datasets/mythlang","meta":{"source":"mythlang_docs","lang":"en"}}'

---

Example Interaction

User:

> How do I declare a function in MythLang?

Bot:

💡 Answer
Functions are declared using the ritual keyword. For example:
ritual functionName(parameters) {
  // function code here
}


You can refer to the "MythLang Syntax" chapter for more details.

📖 Sources
#1 intro.md 0 (score 0.680)
#2 syntax.md 0 (score 0.672)
#3 examples.md 0 (score 0.611)

---

Architecture

.

---

Purpose

This project is not meant as a production-ready system. It’s a portfolio showcase to demonstrate:

• Understanding of RAG pipelines.
• Integration with Telegram bots.
• Handling multiple datasets (real + synthetic).
• Modern MERN + vector DB stack.

---

License

MIT — feel free to fork and adapt.