🤖 ChatVerse AI - Core Backend

A blazingly fast FastAPI intelligent routing engine for text, voice, and multimodal Gemini & Groq with RAG capabilities.

🌐 Overview

The ChatVerse AI Backend is the powerful middleware orchestrating intelligent AI interactions between the React frontend and multiple LLM providers. Built with asynchronous Python using FastAPI and powered by Google Gemini, Groq, and RAG technologies, this server handles API key security, dynamic personas, multimodal processing, and vector-based document retrieval.

Key Mission: Provide a scalable, secure, and intelligent API layer that supports text chat, voice processing, image analysis, and intelligent document retrieval with fallback mechanisms.

🏗️ Architecture

Key Separation:

P2P chat between users is direct Frontend-to-Firebase, bypassing the backend entirely for maximum latency optimization
AI conversations flow through the backend for LLM processing and RAG context retrieval
This decoupled architecture ensures efficient resource utilization and real-time user experiences

� Related Repositories

Frontend: ChatVerse AI Frontend

�🛠️ Backend Technology Stack

System Architecture Diagram

┌─────────────────────────────────────────────────────────────┐
│                   ChatVerse AI Backend                      │
│                    Architecture Stack                       │
└─────────────────────────────────────────────────────────────┘

┌─────────────────────────────────────────────────────────────┐
│  FRONTEND INTERFACE                                         │
│  (React + TypeScript Frontend)                              │
└────────────────────┬────────────────────────────────────────┘
                     │ HTTP/REST
┌────────────────────▼────────────────────────────────────────┐
│  FASTAPI SERVER (Async Python 3.11)                         │
│  ├─ CORS Middleware                                         │
│  ├─ Authentication (Firebase)                               │
│  └─ Request Routing & Validation                            │
└────────────────────┬────────────────────────────────────────┘
                     │
        ┌────────────┼────────────┐
        │            │            │
   ┌────▼──┐  ┌──────▼──┐  ┌──────▼──┐
   │ GEMINI│  │ GROQ    │  │ RAG     │
   │ 2.5   │  │ LLM     │  │ PIPELINE│
   │ Flash │  │ (Fback) │  │         │
   └────┬──┘  └────┬────┘  └────┬────┘
        │          │            │
        └──────────┼────────────┘
                   │
        ┌──────────┴───────────┬─────────────┐
        │                      │             │
   ┌────▼──────┐      ┌───────▼────┐  ┌───▼──────┐
   │ PINECONE  │      │ SUPABASE   │  │ FIREBASE │
   │ Vector DB │      │ File Store │  │ Auth/DB  │
   │ (Embeddings)     │ (Documents)│  │ (Data)   │
   └───────────┘      └────────────┘  └──────────┘
        │                      │             │
        └──────────────────────┴─────────────┘
                   │
            ┌──────▼──────┐
            │ EXTERNAL    │
            │ SERVICES    │
            ├─ Speech2Text│
            ├─ PDF Parser │
            └─ Embeddings │
            └─────────────┘

Tech Stack Details

Component	Technology	Purpose
Framework	FastAPI	High-performance async web framework
Runtime	Python 3.11	Server runtime
LLM #1	Google Gemini 2.5 Flash	Primary AI model with vision capabilities
LLM #2	Groq API	Fallback LLM for chat operations
Vector DB	Pinecone	Semantic search & embeddings storage
RAG Engine	Custom RAG Pipeline	Document-based context retrieval
File Storage	Supabase (PostgreSQL)	Document metadata & file storage
Authentication	Firebase Admin SDK	User verification & JWT validation
Audio Processing	SpeechRecognition	Audio-to-text transcription
PDF Processing	PyPDF2 + pdfplumber	PDF content extraction
Image Processing	Pillow	Image manipulation & analysis
HTTP Client	aiohttp + requests	Async HTTP calls
Embeddings	Gemini API	Text vectorization for RAG
Deployment	Render	Cloud hosting platform

📁 Project Structure

chatverse-ai-backend/
├── main.py                 # FastAPI application & endpoints
├── requirements.txt        # Python dependencies
├── .env                    # Environment variables (not in git)
├── config/
│   ├── config.py          # Configuration loader
│   └── __init__.py
├── auth/
│   ├── firebase_auth.py   # Firebase token verification
│   └── __init__.py
├── services/
│   ├── chat_service.py    # Chat operations
│   ├── rag_service.py     # RAG pipeline
│   ├── document_service.py # Document management
│   ├── data_processor.py  # Content processing
│   ├── embeddings.py      # Vector generation
│   ├── pinecone_handler.py # Vector DB operations
│   ├── supabase_handler.py # File storage operations
│   └── __init__.py
├── test/
│   ├── test_models_google.py
│   ├── test_models_groq.py
│   └── __init__.py
└── README.md              # This file

⚡ Core Features

1. Dual LLM Chat with Intelligent Fallback

Primary: Google Gemini 2.5 Flash with extended context
Fallback: Groq API for redundancy
Dynamic persona support (Assistant, Therapist, Study Buddy, Roast Bot)

2. Retrieval-Augmented Generation (RAG)

Upload documents (PDF, images, audio)
Automatic embedding generation
Semantic search using Pinecone vector database
Context-aware responses based on document content

3. Multimodal Processing

Image Analysis: Upload images for Gemini vision analysis
Voice Transcription: Real-time audio-to-text conversion from .webm files
Multipart Payload Handling: Native binary file processing

4. Document Management

Upload and store documents securely
Retrieve document embeddings from Pinecone
Delete documents with cascading cleanup
User-specific document isolation

5. Conversation Management

Chat history storage in Firestore
Delete specific conversations
Clean history of interactions

6. Firebase Integration

Secure token-based authentication
User session management
Data persistence in Firestore

📡 API Endpoints

Health & System

Method	Endpoint	Description
`GET`	`/health`	Server health check

Chat Endpoints

1. Standard Chat with Fallback

POST /api/chat

Purpose: Direct chat with LLM (Gemini → Groq fallback)
Authentication: No
Request Body:

{
  "message": "Your message here",
  "history": [
    {"sender": "user", "text": "Previous user message"},
    {"sender": "assistant", "text": "Previous AI response"}
  ],
  "persona": "Assistant"
}

Response:

{
  "response": "AI generated response text"
}

2. RAG-Powered Chat

POST /api/rag-chat

Purpose: Query with document context retrieval
Authentication: Firebase Token Required ✅
Request Body:

{
  "query": "What is mentioned about topic X in my documents?"
}

Response:

{
  "response": "Context-aware response from RAG pipeline",
  "sources": [
    {
      "document_id": "doc_123",
      "filename": "document.pdf",
      "relevance": 0.95
    }
  ]
}

Document Management Endpoints

3. Upload Document

POST /api/upload-document

Purpose: Upload document for RAG indexing
Authentication: Firebase Token Required ✅
Content-Type: multipart/form-data
Request Parameters:

file (File): PDF, image, or audio file
Auto-extracted: user_id from Firebase token

Response:

{
  "success": true,
  "filename": "document.pdf",
  "message": "File is being processed. It will be available for queries shortly."
}

4. Get User Documents

GET /api/documents

Purpose: Retrieve all documents uploaded by user
Authentication: Firebase Token Required ✅

Response:

{
  "documents": [
    {
      "doc_id": "doc_123",
      "filename": "research.pdf",
      "upload_date": "2024-01-15T10:30:00Z",
      "file_size": 2048576,
      "status": "processed"
    }
  ]
}

5. Delete Specific Document

DELETE /api/documents/{doc_id}

Purpose: Delete single document from all storage systems
Authentication: Firebase Token Required ✅
Path Parameters:

doc_id: Document identifier

Response:

{
  "success": true,
  "message": "Document deleted successfully"
}

6. Delete All Documents

DELETE /api/documents/delete-all

Purpose: Clear all documents for authenticated user
Authentication: Firebase Token Required ✅

Response:

{
  "success": true,
  "message": "All documents deleted successfully"
}

Image & Vision Endpoints

7. Image Analysis

POST /api/image-scan

Purpose: Analyze image with Gemini Vision
Authentication: No
Content-Type: multipart/form-data
Request Parameters:

file (File): Image file (.jpg, .png, etc.)
prompt (string): Analysis prompt

Response:

{
  "response": "Detailed image analysis from Gemini"
}

Audio Processing Endpoints

8. Voice Transcription

POST /api/voice

Purpose: Convert audio to text
Authentication: No
Content-Type: multipart/form-data
Request Parameters:

file (File): Audio file (.webm, .mp3, .wav, etc.)

Response:

{
  "success": true,
  "transcript": "Transcribed text from audio",
  "filename": "recording.webm"
}

Conversation Management Endpoints

9. Delete Chat Conversation

DELETE /api/chat/{conversation_id}

Purpose: Delete specific conversation and all messages
Authentication: Firebase Token Required ✅
Path Parameters:

conversation_id: Chat type (e.g., 'assistant', 'rag-analysis', 'therapist')

Response:

{
  "success": true,
  "message": "Chat conversation 'assistant' deleted successfully"
}

📦 Core Services

1. RAG Service (`rag_service.py`)

Handles intelligent document retrieval and context augmentation:

Query embedding generation
Semantic similarity search in Pinecone
Context ranking and selection
Fallback handling between Gemini and Groq

2. Document Service (`document_service.py`)

Manages document lifecycle:

Upload processing
Storage in Supabase
Metadata in Firestore
Cascade deletion across systems

3. Chat Service (`chat_service.py`)

Conversation management:

Message persistence
History retrieval
Conversation deletion
User session tracking

4. Data Processor (`data_processor.py`)

Multimodal content processing:

PDF text extraction
Image preprocessing
Audio transcription
Data normalization

5. Embeddings (`embeddings.py`)

Vector generation:

Gemini-powered text embeddings
Chunk-based processing
Embedding caching

6. Pinecone Handler (`pinecone_handler.py`)

Vector database operations:

Embedding storage
Semantic search
Vector deletion

7. Supabase Handler (`supabase_handler.py`)

File and data storage:

Document file upload
Metadata persistence
File retrieval

🚀 Setup & Installation

Prerequisites

Python 3.11+
pip package manager
Git

Step 1: Clone Repository

git clone https://github.com/Rahul-8283/chatverse-ai-backend.git
cd chatverse-ai-backend

Step 2: Create Virtual Environment

# Windows
python -m venv venv
venv\Scripts\activate

# macOS/Linux
python3 -m venv venv
source venv/bin/activate

Step 3: Install Dependencies

pip install -r requirements.txt

Step 4: Configure Environment Variables

Create a .env file in the root directory:

GEMINI_API_KEY=your_gemini_api_key
GROQ_API_KEY=your_groq_api_key

PINECONE_INDEX_NAME=your_pinecone_index_name
PINECONE_API_KEY=your_pinecone_api_key

SUPABASE_URL=your_supabase_url
SUPABASE_BUCKET=your_supabase_bucket_name
SUPABASE_SECRET_KEY=your_supabase_secret_key

FIREBASE_PROJECT_ID=project_id
FIREBASE_CLIENT_ID=your_firebase_client_id
FIREBASE_CLIENT_EMAIL=your_firebase_client_email
FIREBASE_PRIVATE_KEY=your_firebase_private_key
FIREBASE_PRIVATE_KEY_ID=your_firebase_private_key_id

Step 5: Download Firebase Credentials

Go to Firebase Console
Navigate to Project Settings → Service Accounts
Click "Generate New Private Key"
Save as firebase-service-account-key.json in project root

Step 6: Run Development Server

uvicorn main:app --reload --host 0.0.0.0 --port 8000

Access API Documentation:

Swagger UI: http://localhost:8000/docs
ReDoc: http://localhost:8000/redoc

📦 Dependencies Overview

Core Framework:
  ├─ FastAPI (Web framework)
  ├─ Uvicorn (ASGI server)
  └─ Python-multipart (Form data parsing)

AI/ML:
  ├─ google-generativeai (Gemini API)
  ├─ groq (Groq LLM API)
  ├─ pinecone-client (Vector DB)
  └─ numpy (Numerical computing)

Data Processing:
  ├─ PyPDF2 (PDF reading)
  ├─ pdfplumber (PDF extraction)
  ├─ Pillow (Image processing)
  └─ pandas (Data manipulation)

Cloud Services:
  ├─ firebase-admin (Firebase integration)
  ├─ supabase (Backend as a Service)
  └─ requests (HTTP client)

Audio:
  └─ SpeechRecognition (Audio transcription)

📄 License

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

FilesExpand file tree

README.md

Latest commit

History

README.md

File metadata and controls

🤖 ChatVerse AI - Core Backend

🌐 Overview

🏗️ Architecture

� Related Repositories

�🛠️ Backend Technology Stack

System Architecture Diagram

Tech Stack Details

📁 Project Structure

⚡ Core Features

1. Dual LLM Chat with Intelligent Fallback

2. Retrieval-Augmented Generation (RAG)

3. Multimodal Processing

4. Document Management

5. Conversation Management

6. Firebase Integration

📡 API Endpoints

Health & System

Chat Endpoints

1. Standard Chat with Fallback

2. RAG-Powered Chat

Document Management Endpoints

3. Upload Document

4. Get User Documents

5. Delete Specific Document

6. Delete All Documents

Image & Vision Endpoints

7. Image Analysis

Audio Processing Endpoints

8. Voice Transcription

Conversation Management Endpoints

9. Delete Chat Conversation

📦 Core Services

1. RAG Service (rag_service.py)

2. Document Service (document_service.py)

3. Chat Service (chat_service.py)

4. Data Processor (data_processor.py)

5. Embeddings (embeddings.py)

6. Pinecone Handler (pinecone_handler.py)

7. Supabase Handler (supabase_handler.py)

🚀 Setup & Installation

Prerequisites

Step 1: Clone Repository

Step 2: Create Virtual Environment

Step 3: Install Dependencies

Step 4: Configure Environment Variables

Step 5: Download Firebase Credentials

Step 6: Run Development Server

📦 Dependencies Overview

📄 License

1. RAG Service (`rag_service.py`)

2. Document Service (`document_service.py`)

3. Chat Service (`chat_service.py`)

4. Data Processor (`data_processor.py`)

5. Embeddings (`embeddings.py`)

6. Pinecone Handler (`pinecone_handler.py`)

7. Supabase Handler (`supabase_handler.py`)