Decentralized AI Simulator

We are developing a scalable framework-agnostic decentralized AI platform for research community.

View Demo · Documentation · Report Bug · Request Feature

📔 Table of Contents

• About the Project
  • Screenshots
  • Tech Stack
  • Features
  • Color Reference
  • Environment Variables
• Getting Started
  • Run Locally
• Roadmap
• Contributing
  • Code of Conduct
• FAQ
• License
• Contact
• Acknowledgements

🌟 About the Project

📷 Screenshots

👾 Tech Stack

• Ubuntu
• Conda
• Python 10+
• Pytorch
• Torch Vision
• Node
• gRPC
• Docker
• Kubernetes
• GitHub
• React/Typescript/Javascript

🎯 Features (planned)

• 🤔 Framework-agnostic --> Bring any kind of ML/DL/AI framework from Pytorch and TensorFlow to Keras and Vanilla ML libraries.

• 🤔 Decentralized --> No single point of failure in the entire system. All agents can communicate without centralized control.

• 🤔 Scalable --> The platform supports every kind of data source and every kind of compute device/system

• 🤔 Secure --> The system is designed with security-first approach

• 🤔 Privacy-preserving --> The privacy of each data producer and data subject is preserved across the system

• 🤔 Device-Native --> The control and preferences lies at the device-end

• 🤔 Distributed --> The platform complies with all functional and non-functional requirements of a distributed system

🔑 Environment

To run this project, you will need to create a virtual environment and install

• Python 3.10+
• torch
• torch vision

🧰 Getting Started

🏃 Run Locally

Clone the project

  git clone https://github.com/mhrehman17/decentai.git

Go to the download location, and run following command.

  python -m decentai.main 

Now you can see an MNIST example running on screen.

🧭 Roadmap

• Local multi-agent simulator. You can configure as many training and evaluation agents as you like to.
• MNIST example
• CIFAR 10 example
• Multiple aggregators
• Multiple metrics
• Directory structure and stub files for data pipelines
• Directory structure and stub files for differential privacy
• Documentation folder added
• Directory structure and stub files for homomorphic encryption
• SMPC
• Device-agnostic gRPC peer-to-peer communication model
• Dockerisation/Virtualisation/Kubernetes
• Cloud/Edge/On-Prem Deployment
• CI/CD
• GitHub Actions

👋 Contributing

Contributions are always welcome! Please feel free to submit pull request, if you want to propose new features, or submit an issue.

See contributing.md for ways to get started.

📜 Code of Conduct

Please read the Code of Conduct

❔ FAQ (coming soon)

• Question 1

  • Answer 1

• Question 2

  • Answer 2

⚠️ License

Distributed under the no License. See LICENSE.txt for more information.

🤝 Contact

Habib Rehman - @habibcomsats - [email protected]

Project Link: https://github.com/mhrehman17/decentai

💎 Acknowledgements

The project is a collective effort of researchers from LEADS and its collaborators.