Does FL miss a coordination layer? Would coordination layer make it better to scale FL?

[Swayam1000]

I’m currently building SoraChain AI — a coordination layer for Federated Learning, designed to help individuals, researchers, and institutions collaborate globally in a trustless, privacy-preserving way.

I’ve long believed that FL is the right paradigm to build AI that truly serves humanity — but scaling it beyond a single organization remains incredibly challenging. Trust, governance, incentives, and reproducibility are still very much open problems.

We’re eager to understand these challenges more deeply, and input from practitioners like you is essential to ensure we’re solving real, high-leverage problems — and helping FL take a step closer to its AlphaGo moment.

If you’re working with FL systems, your insights would be incredibly valuable in shaping what we’re building at SoraChain AI.

Survey form: https://tally.so/r/3yrZd4

If you’re curious to learn more, I’m happy to answer questions here.

[chesterxgchen]

Your link doesn't exists. What's your definition for coordination layer? Sounds like a shared FL platform allow everyone to collaborate. You have listed the challenges and missing the legal and approval.

[SoraChainAI]

How do actors, such as Hospital A (London) and Hospital B (Sydney), collaborate to train a model together without sharing raw data? Our hypothesis is that a blockchain embedded as a trustless layer could enable this collaboration.

Would appreciate if you can participate inthis short survey: ttps://tally.so/r/3yrZd4