GreenSync

Landing Page
Chat window
Dashboard
Donation Page

Inspiration

With millions using AI like ChatGPT daily, the environmental impact of large-scale computation is immense. We wanted to create a sustainable alternative that not only reduces reliance on energy-hungry data centers but also rewards users for contributing computing power. Our goal was to make AI usage more eco-friendly and measurable, giving users insight into their carbon footprint reduction while enabling decentralized AI processing.

What It Does

GreenSync distributes AI computation across multiple devices, reducing the need for centralized cloud servers. Users receive Green Coins based on their computational contributions, incentivizing participation in a more sustainable AI ecosystem. The platform can also scale up large AI models across many users’ devices, making it possible to run powerful models that would typically require expensive cloud resources.

How We Built It

We used FastAPI for the backend, handling authentication, credit calculations, and API endpoints. Ray was implemented to distribute computation efficiently across multiple devices. The frontend, built with React and Tailwind CSS, provides users with real-time feedback on their Green Coin balance and environmental impact.

Unlike traditional AI models that depend on centralized data centers, GreenSync runs entirely on a decentralized network of user devices. Instead of offloading computation to the cloud, we built an efficient resource-sharing system that dynamically distributes AI tasks among connected devices.

Challenges We Ran Into

Running computation directly in the browser was difficult due to security and performance constraints.
Ensuring smooth AI task distribution across devices while avoiding overload or inefficiencies.
Managing real-time updates of Green Coins and credits between the dashboard and donation pages to reflect accurate values.
Optimizing Ray's task scheduling to maximize efficiency across different hardware configurations.

Accomplishments That We're Proud Of

Successfully distributed AI processing across multiple user devices without relying on cloud infrastructure.
Developed a real-time credit system that measures user contributions and rewards them fairly.
Built an engaging, user-friendly interface that tracks Green Coin earnings and environmental impact.
Overcame technical hurdles with Ray and FastAPI to scale AI workloads efficiently.

What We Learned

Decentralized AI is possible but challenging due to browser limitations.
Optimizing computation sharing between devices is crucial for performance.
Efficient task scheduling with Ray is essential for optimal resource utilization.
Real-time UI updates require careful state management between pages.

What's Next for GreenSync

Expanding support for more AI models that can be efficiently distributed across devices.
Improving task allocation algorithms to optimize resource usage in real time.
Allowing users to pool computation for even larger AI models.
Partnering with sustainability initiatives to convert Green Coins into real-world impact*, such as funding **carbon offset projects*.