Artifex

Artifex addresses the underutilization of retail GPUs in the POS era by utilizing them to democratize AI. The advent of complex AI models such as stable diffusion requires substantial computing power for inference tasks, which presents the use of idle GPUs to power them. This provides a win-win scenario for the owner of the idle GPUs, as it earns them money, and the people who make use of the extra computing power, as it allows them to use more computationally intensive AI services. With a focus on democratizing AI, Artifex offers a scalable, accessible, and decentralized platform for efficient utilization of AI models. Through stable diffusion, Artifex assures high performance, robustness, and scalability of AI services, catering to a wide array of use cases. Not only does Artifex mobilize idle GPU resources, but it also opens up opportunities for AI model creators to monetize their work. By doing so, Artifex is turning the challenge of limited computing power into an opportunity for increased efficiency and innovation in the AI landscape.

This project is developed with hardhat and we used React for the frontend. In the backend we used stable-diffusion-web-ui to perform inference and use web3.py to listen to smart contract event and call smart contract functions. The result returned by backend is permanently stored to IPFS and the cid to the object is broadcasted as an event, which will be picked by frontend so that result can be rendered to end users.