---
title: RAG Research Assistant API
emoji: 📚
colorFrom: blue
colorTo: green
sdk: docker
app_port: 7860
pinned: false
---

# RAG Research Assistant API

This backend provides a sophisticated FastAPI-based service for a Retrieval Augmented Generation (RAG) system designed to assist with research paper searches and information retrieval.

## Features

- ArXiv paper search with customizable filtering
- Document chunking and processing
- Embedding generation using Sentence Transformers 
- Vector search using FAISS
- LLM response generation with multiple model options
- Markdown-formatted research results

## Requirements

- Python 3.8+
- FastAPI
- Sentence Transformers
- FAISS
- Hugging Face API access

## Installation

1. Clone the repository
2. Navigate to the backend directory
3. Install dependencies:

```bash
pip install -r requirements.txt
```

4. Create a `.env` file based on `.env.example` and add your Hugging Face API key

## Running the Application

To run the development server:

```bash
python run.py
```

Or with uvicorn directly:

```bash
uvicorn app.main:app --reload
```

The API will be available at http://localhost:8000, and the API documentation at http://localhost:8000/docs.

## API Endpoints

- `/rag/query`: Process a query through the RAG pipeline
- `/rag/search`: Search for papers without LLM processing
- `/rag/models`: Get available LLM models
- `/rag/stats`: Get system statistics
- `/rag/clear/cache`: Clear the paper cache
- `/rag/clear/database`: Clear the vector database
- `/health`: Simple health check endpoint

## Docker

You can also run the application using Docker:

```bash
docker build -t rag-backend .
docker run -p 8000:8000 -e HF_API_KEY=your_key_here rag-backend
```

## Architecture

The application follows a service-oriented architecture:

- `ArxivService`: Interface to ArXiv API
- `DocumentService`: Process papers into chunks
- `EmbeddingService`: Generate embeddings
- `VectorService`: Store and search vectors
- `LlmService`: Generate responses using LLMs
- `FormatterService`: Format results
- `RagService`: Orchestrate the entire RAG pipeline