Your Hardware, Their Intelligence, Our Network.
Aight is a hackathon-grade DePIN marketplace for local LLM inference. Operators stake native ETH on Base Sepolia, expose an Ollama-backed endpoint, and serve OpenAI-compatible requests through the Aight Gateway. Users lock prepaid inference hours in escrow and receive an AIGHT_API_KEY for Cursor, OpenClaw, or any OpenAI-compatible client.
User Client -> Aight Gateway -> Operator Tunnel -> Local Ollama
| | |
| | v
| | Operator Hardware
v v
Base Sepolia <- Telemetry WebSocket <- Pulse Dashboard
contracts/ Foundry workspace for AightRegistry staking and escrow
gateway/ FastAPI and LiteLLM OpenAI-compatible proxy
operator/ Local operator CLI workspace
frontend/ Next.js Pulse dashboard workspace
contracts/contains the native ETHAightRegistrycontract, Foundry tests, Base Sepolia config, and deployment script.gateway/contains the FastAPI proxy scaffold, dummy operator endpoint, API-key issuance flow, LiteLLM routing, and WebSocket telemetry.operator/andfrontend/are reserved for the next implementation phases.
cd contracts
forge install OpenZeppelin/openzeppelin-contracts foundry-rs/forge-std
Copy-Item .env.example .env
forge testDeploy to Base Sepolia:
forge script script/DeployAightRegistry.s.sol:DeployAightRegistry --rpc-url base_sepolia --broadcast --verifycd gateway
python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt
Copy-Item .env.example .env
uvicorn main:app --reload --port 8787The gateway exposes:
POST /admin/operatorsto register a hackathon operator endpoint.POST /admin/api-keysto issue anAIGHT_API_KEYfor a funded escrow.POST /v1/chat/completionsfor OpenAI-compatible inference.WS /ws/telemetryfor token flow events.POST /dummy/v1/chat/completionsfor local proxy testing.