Skip to content

Repository files navigation

:) OBSIZEN

Transform your Obsidian vault into an intelligent AI knowledge companion.

Your second brain just got a search engine and an AI assistant.

✨ Features🏗 Architecture🚀 Getting Started📖 Usage🗺 Roadmap


:O What is OBSIZEN?

OBSIZEN is an open-source Retrieval-Augmented Generation (RAG) system built specifically for Obsidian users.

Instead of manually digging through hundreds or thousands of notes, OBSIZEN turns your vault into an intelligent knowledge companion that can:

  • 🔍 Search semantically across your notes
  • 📄 Understand filenames and metadata
  • 🧩 Combine multiple retrieval strategies
  • 🎯 Rerank results for better answers
  • 🤖 Generate grounded responses using LLMs

Think of it as:

ChatGPT for your Obsidian vault, except it actually knows your notes.


:D Features

🔎 Hybrid Retrieval

Combines:

  • Vector Search
  • BM25 Search
  • Filename Search

🏷 Metadata Awareness

Understands:

  • Tags
  • Frontmatter
  • File structure
  • Document context

;) Fast & Efficient

  • Incremental indexing
  • Cached models
  • Reduced startup time

:} Better Answers

  • Source deduplication
  • CrossEncoder reranking
  • Grounded generation

📸 Demo

OBSIZEN Demo

🏗 Architecture

                ┌────────────────┐
                │ Obsidian Vault │
                └────────┬───────┘
                         │
                         ▼
               ┌──────────────────┐
               │ Markdown Loader  │
               └────────┬─────────┘
                        │
                        ▼
          ┌───────────────────────────┐
          │ Markdown Header Chunking  │
          └────────┬──────────────────┘
                   │
                   ▼
          ┌───────────────────────────┐
          │ Embedding Generation      │
          └────────┬──────────────────┘
                   │
                   ▼
             ┌─────────────┐
             │  ChromaDB   │
             └──────┬──────┘
                    │
                    ▼
     ┌───────────────────────────────────┐
     │ Hybrid Retrieval Engine           │
     │ Vector + BM25 + Filename Search   │
     └───────────────────────────────────┘
                    │
                    ▼
          ┌────────────────────┐
          │ Metadata Boosting  │
          └────────┬───────────┘
                   │
                   ▼
         ┌─────────────────────┐
         │ Source Deduplication│
         └────────┬────────────┘
                  │
                  ▼
         ┌─────────────────────┐
         │ CrossEncoder Ranker │
         └────────┬────────────┘
                  │
                  ▼
         ┌─────────────────────┐
         │   Context Builder   │
         └────────┬────────────┘
                  │
                  ▼
       ┌──────────────────────────┐
       │ DeepSeek via OpenRouter  │
       └───────────┬──────────────┘
                   │
                   ▼
          🌌 Grounded AI Answer

Getting Started

Clone the Repository

git clone https://github.com/sudeepmukul/obsizen.git
cd obsizen

Install Dependencies

pip install -r requirements.txt

Configure

Create your environment file:

OPENROUTER_API_KEY=your_api_key
OBSIDIAN_VAULT_PATH=/path/to/your/vault

📖 Usage

Build the Index

python src/main.py

Ask Questions

What are my YC ideas?
What did I learn about FastAPI?
Summarize my ML notes.
What projects am I procrastinating on?

OBSIZEN retrieves relevant notes, reranks the results, and generates a grounded answer directly from your vault.


⚙️ Tech Stack

LangChain
ChromaDB
SentenceTransformers
CrossEncoder
OpenRouter
Python

🗺 Roadmap

  • Hybrid Retrieval
  • Metadata Retrieval
  • Incremental Indexing
  • CrossEncoder Reranking
  • Logging System
  • Comprehensive Tests
  • Better Error Handling
  • FastAPI Backend
  • React Web UI
  • Local LLM Support
  • Docker Deployment
  • Obsidian Plugin

🤝 Contributing

Contributions are welcome and greatly appreciated.

Fork 
Clone 
Create a Branch 
Commit 
Open a Pull Request 

If you have ideas, feature requests, or improvements, open an issue and let's build a better second brain together.


⭐ Support the Project

If OBSIZEN helped you organize your knowledge and think better:

🌟 Star the repository 🐛 Report bugs 💡 Suggest features 🤝 Contribute code


🌌 Built with <3 by Sudeep Mukul & Zency R&D Team

Transform your notes into an AI-powered second brain.

If this project helped you, consider giving it a ⭐

About

Advanced Retrieval-Augmented Generation (RAG) system for chatting with an Obsidian knowledge base using LangChain, ChromaDB, Hugging Face embeddings, CUDA acceleration , Hybrid Search and Reranking

Topics

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages