MHamdan/SPARKNET

🔥 SPARKNET: AI-Powered Technology Transfer Office Automation

Multi-agent AI platform for research valorization and IP management

  

Overview

SPARKNET is an enterprise-grade Technology Transfer Office (TTO) Automation Platform that combines multi-agent AI orchestration with document intelligence to automate key TTO workflows. Built for the VISTA/Horizon EU project.

🎯 Core TTO Scenarios

📊 Coverage Dashboard

• —3 Fully Covered - Production-ready scenarios
• —5 Partially Covered - In development
• —2 Not Covered - Planned for future

Features

🛡️ AI Quality Assurance

• —CriticAgent Validation: Every AI output validated against VISTA quality standards
• —Confidence Scoring: Automatic abstention for low-confidence results
• —Source Verification: Hallucination mitigation with evidence grounding
• —Human-in-the-Loop: Critical decisions require human approval

🤖 Multi-Agent Architecture

• —PlannerAgent: Task decomposition and workflow planning
• —ExecutorAgent: Task execution with tool usage
• —CriticAgent: Output validation and refinement
• —MemoryAgent: Context management and retrieval

📄 Document Intelligence

• —OCR with PaddleOCR/Tesseract
• —Layout detection and semantic chunking
• —Schema-driven field extraction
• —Visual evidence grounding (bbox, page, confidence)

💬 RAG Q&A

• —Vector search with ChromaDB
• —Grounded retrieval with citations
• —Multi-document querying
• —Citation generation

Quick Start

Streamlit Cloud (Recommended)

The app is deployed on Streamlit Cloud. Visit:

https://sparknet.streamlit.app

Local Installation

# Clone repository
git clone https://github.com/MHHamdan/SPARKNET.git
cd SPARKNET

# Install dependencies
pip install -r requirements.txt

# Run Streamlit app
streamlit run demo/app.py

With Local LLM (Ollama)

For privacy-preserving local inference:

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# Pull models
ollama pull llama3.2:latest
ollama pull nomic-embed-text

# Run SPARKNET
streamlit run demo/app.py

Configuration

API Keys

Configure in .streamlit/secrets.toml or environment variables:

[auth]
password = "your-password"

GROQ_API_KEY = "your-groq-key"
GOOGLE_API_KEY = "your-google-key"
OPENROUTER_API_KEY = "your-openrouter-key"

See .env.example for all available configuration options.

Supported LLM Providers

Project Structure

SPARKNET/
├── demo/                    # Streamlit application
│   ├── app.py              # Main app
│   ├── auth.py             # Authentication
│   ├── llm_providers.py    # LLM provider management
│   └── pages/              # Multi-page app
├── src/
│   ├── agents/             # Agent implementations
│   │   ├── scenario1/      # Patent Wake-Up
│   │   ├── scenario3/      # License Compliance
│   │   └── scenario4/      # Award Identification
│   ├── rag/                # RAG subsystem
│   ├── workflow/           # LangGraph workflows
│   └── document_intelligence/  # Document processing
├── configs/                # Configuration files
├── .streamlit/             # Streamlit config
└── SECURITY.md            # Security documentation

Security & GDPR

SPARKNET supports GDPR-compliant deployments:

• —Local Inference: Use Ollama for 100% on-premise processing
• —Data Isolation: Configure data retention policies
• —Audit Logging: Track all AI interactions
• —Private Deployment: Enterprise deployment options

See SECURITY.md for detailed security documentation.

Development

Running Tests

pytest tests/

Code Formatting

black src/
flake8 src/

Roadmap

• —[x] Patent Wake-Up workflow
• —[x] Agreement Safety review
• —[x] Partner Matching
• —[x] CriticAgent validation
• —[ ] License Compliance Monitoring (in progress)
• —[ ] Award Identification (in progress)
• —[ ] Grant Writing Assistant
• —[ ] Negotiation Support

Contributing

Contributions are welcome! Please:

1. 1.Fork the repository
2. 2.Create a feature branch
3. 3.Make your changes
4. 4.Run tests
5. 5.Submit a pull request

Acknowledgments

• —Ollama for local LLM inference
• —NVIDIA for CUDA and GPU support
• —LangChain for LLM orchestration
• —Streamlit for the web framework
• —The open-source AI community

Support

• —GitHub Issues: github.com/MHHamdan/SPARKNET/issues
• —Documentation: See /docs folder

License

MIT License - see LICENSE file for details.

<p align="center"> <strong>🔥 SPARKNET</strong><br> AI-Powered Technology Transfer Office Automation<br> <em>VISTA/Horizon EU Project</em> </p>