🍜 NeuralNoodle

NeuralNoodle is a production-grade, agentic Retrieval-Augmented Generation (RAG) system designed for deep technical investigations. It behaves like an expert engineer—forming hypotheses, planning investigations, and synthesizing complex information across vast document corpora.

---

✨ Key Features

• 🧠 Agentic Reasoning: Built on LangGraph, the system doesn't just search; it investigates by forming and testing hypotheses.
• 🚀 High-Density Workspace: A professional-grade UI for managing thousands of documents with real-time pipeline visualization.
• 🔍 Hybrid Retrieval: Combines Dense (Gemini Embeddings) and Sparse (SPLADE/BM25) search for maximum precision and recall.
• 🛡️ Evidence Synthesis: Automatically clusters supporting and contradicting evidence into a coherent knowledge graph.
• ⚡ NeuralNoodle Pipeline: A high-performance ingestion engine with neuromorphic mapping and real-time logging.

---

🏗️ System Architecture

NeuralNoodle uses a multi-node agentic flow to simulate expert human investigation:

graph TD
    A[User Query] --> B[Hypothesis Generator]
    B --> C[Task Planner]
    C --> D[Hybrid Retrieval]
    D --> E[Cross-Encoder Reranker]
    E --> F[Evidence Evaluator]
    F --> G{Confidence?}
    G -- No --> H[Query Refiner]
    H --> D
    G -- Yes --> I[Synthesis Node]
    I --> J[Final Answer]

Loading

---

🛠️ Quick Start

1. Prerequisites

• Python 3.11+
• Node.js & npm (for the frontend)
• Docker (for Qdrant vector store)
• Google API Key (for Gemini)

2. Installation

# Clone the repo
git clone https://github.com/yourusername/NeuralNoodle.git
cd NeuralNoodle

# Backend Setup
poetry install
cp .env.example .env  # Add your API keys

# Frontend Setup
cd frontend
npm install

3. Launching the System

# Start Qdrant
docker-compose up -d qdrant

# Run the Backend
python src/main.py

# Run the Frontend
cd frontend
npm run dev

---

📂 Repository Structure

• src/: Backend logic (Agents, Ingestion, Retrieval).
• frontend/: React/Vite dashboard.
• docs/: Design documents, product vision, and architecture details.
• config/: System and model configurations.
• tests/: Comprehensive test suite.
• data/: Local storage for uploads and processed shards (Git ignored).

---

📄 License

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

---

Built with ❤️ for the future of knowledge engineering.