Light - Semantic Document Search Engine

A powerful semantic document search engine built with Next.js, FastAPI, Ollama, and ChromaDB. Generate AI-powered documents and search through them using natural language queries with vector embeddings.

Features

• 🤖 AI Document Generation: Generate documents using Ollama's GPT-OSS model
• 🔍 Semantic Search: Search documents using natural language with vector embeddings
• 💾 Vector Storage: ChromaDB for efficient vector storage and retrieval
• 🎨 Premium UI: Beautiful dark mode interface with glassmorphism effects
• ⚡ Real-time: Instant search results with semantic similarity scoring
• 📱 Responsive: Works seamlessly on desktop and mobile devices

Tech Stack

Frontend

• Next.js 15 - React framework with App Router
• TypeScript - Type-safe development
• Tailwind CSS - Utility-first styling
• Shadcn UI - Premium component library
• Framer Motion - Smooth animations
• Bun - Fast package manager and runtime

Backend

• FastAPI - High-performance Python API framework
• Ollama - Local LLM inference
  • gpt-oss:20b - Text generation
  • nomic-embed-text - Vector embeddings
• ChromaDB - Vector database for semantic search
• Uvicorn - ASGI server

Project Structure

light/
├── apps/
│   ├── web/              # Next.js frontend
│   │   ├── src/
│   │   │   ├── app/      # App router pages
│   │   │   ├── components/
│   │   │   │   └── ui/   # UI components
│   │   │   └── lib/      # Utilities
│   │   └── package.json
│   └── backend/          # FastAPI backend
│       ├── main.py       # API endpoints
│       ├── requirements.txt
│       └── chroma_db/    # Vector database storage
└── packages/
    └── ...               # Shared packages

Getting Started

Prerequisites

• Bun (v1.0+) - Install Bun
• Python (v3.10+)
• Ollama - Install Ollama

Installation

1. Clone the repository

   git clone <repository-url>
   cd light

2. Install Ollama models

   ollama pull gpt-oss:20b
   ollama pull nomic-embed-text

3. Install frontend dependencies

   bun install

4. Set up backend

   cd apps/backend
   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate
   pip install -r requirements.txt

Running the Application

1. Start Ollama (if not running)

   ollama serve

2. Start the backend (in apps/backend)

   source .venv/bin/activate
   uvicorn main:app --reload --port 8000

3. Start the frontend (in project root)

   bun run dev

4. Access the application

   • Frontend: http://localhost:3000
   • Backend API: http://localhost:8000
   • API Docs: http://localhost:8000/docs

Usage

Generate Documents

1. Type a prompt in the main input box
2. Click the send button (arrow icon)
3. View the generated document below

Search Documents

1. Press Cmd+K (or Ctrl+K) or click the Search button in the navbar
2. Type your search query
3. Click on any result to view full details

Keyboard Shortcuts

• Cmd/Ctrl + K - Open search dialog
• Enter - Submit prompt or search
• Esc - Close dialogs

API Endpoints

POST /generate

Generate a new document and store it in the vector database.

Request:

{
  "prompt": "Write a haiku about coding"
}

Response:

{
  "id": "uuid-string",
  "prompt": "Write a haiku about coding",
  "content": "Silent keys tap code,\nLogic blooms in glowing lines,\nDawn breaks, bugs vanish."
}

POST /search

Search for documents using semantic similarity.

Request:

{
  "query": "poems about programming"
}

Response:

{
  "results": [
    {
      "id": "uuid-string",
      "content": "document content...",
      "metadata": { "prompt": "original prompt" },
      "distance": 0.123
    }
  ]
}

Development

Build

bun run build

Lint

bun run lint

Type Check

bun run check-types

Configuration

Ollama Models

Edit apps/backend/main.py to change models:

GENERATION_MODEL = "gpt-oss:20b"
EMBEDDING_MODEL = "nomic-embed-text"

ChromaDB Storage

Vector database is stored in apps/backend/chroma_db/

Troubleshooting

Ollama 404 Error

Ensure Ollama is running and models are pulled:

ollama serve
ollama pull gpt-oss:20b
ollama pull nomic-embed-text

Backend Connection Error

Check if the backend is running on port 8000:

curl http://localhost:8000

Frontend Build Issues

Clear cache and reinstall:

rm -rf node_modules bun.lock
bun install

License

MIT

Contributing

Contributions are welcome! Please open an issue or submit a pull request.