Protocol

A peer-to-peer compute protocol for decentralized AI development at scale. This protocol implementation powered distributed training runs including synthetic-2 and intellect-2.

This repository contains the core infrastructure for coordinating compute resources across a distributed network, including workers, validators, and the orchestration layer.

📚 Table of Contents

• System Architecture
• Getting Started
• Documentation
• Frequently Asked Questions
• Community
• Contributing
• Security
• License

System Architecture

The protocol follows a modular architecture designed for decentralized AI compute:

Component Overview

• Smart Contracts: Ethereum-based contracts manage the protocol's economic layer
• Discovery Service: Enables secure peer discovery and metadata sharing
• Orchestrator: Coordinates compute jobs across worker nodes
• Validator Network: Ensures quality through random challenges
• Worker Nodes: Execute AI workloads in secure containers

Getting Started

Prerequisites

• Linux operating system
• CUDA-capable GPU(s) for worker operations
• Docker (version 28.1.1 or later) and Docker Compose (version v2.35.1 or later)

For complete setup instructions, refer to our Development Setup Guide.

Install Worker CLI:

You can install the latest worker CLI using:

curl -sSL https://raw.githubusercontent.com/PrimeIntellect-ai/protocol/main/crates/worker/scripts/install.sh | bash 

This can also be used to upgrade the current installation to the latest release.

For the latest dev build use:

curl -sSL https://raw.githubusercontent.com/PrimeIntellect-ai/protocol/develop/crates/worker/scripts/install.sh | bash -s -- --dev

Documentation

• Development Setup Guide - Detailed installation and environment setup instructions
• Usage Guide - Instructions for dispatching tasks, monitoring, and system management

Frequently Asked Questions

Q: What is this protocol?

A: This is a peer-to-peer compute protocol implementation for decentralized AI development at scale. It provides infrastructure for coordinating compute resources across a distributed network through workers, validators, and an orchestration layer. This protocol powered distributed training runs including synthetic-2 and intellect-2.

Q: Can I run this protocol locally?

A: Yes, you can run the protocol locally for development and testing purposes.

Q: What environment variables do I need for local development?

A: We have provided an .env.example file with the required variables.

Q: How are private keys managed securely in the system?

A: We're actively developing our security practices for private key management.

Q: What are the recommended network isolation strategies for the worker?

A: We will be providing detailed documentation on how to secure workers with firewalls for ingress / egress

Q: What are the funding requirements for workers?

A: For the current development phase, minimal testnet ETH is sufficient. Detailed information on setting up the worker will follow.

Community

• Discord
• X
• Blog

Contributing

We welcome contributions! Please see our Contributing Guidelines.

Security

See SECURITY.md for security policies and reporting vulnerabilities.