P2P Data fabric for incentivized creation of AI over edge compute
Funding Area: Critical shared infrastructure / Infraestructura compartida critica
Problem Statement: AI models and innovation cannot be monopolized. Individuals that generate data by which it feeds need to be compensated
Proposed Activities: think https://www.waterlily.ai/ but with kamu that can merge data sets into new derivative datasets, with compensation information attached to contributor data. models are trained over edge compute with bacalhau, and they are made available online for a fee which is kicked back to the model creators. I could use 6months for an initial prototype and the minimum of grant funding
Openness: all work will be opensource in a public github repo that is free to use. The tech stack is relevant to the filecoin ecosystem and as LLMs become more native in all workflows, all of them
Challenges: distributed computing has its own challenges, this project hinges on Kamu and Bacalhau so its technological limitations will be grounded here
Neglectedness: I'm sure there is, this is my first grant attempt
Success: If we can get a group of writers creating databases with their compensation information, their databases are merged with Kamu and their AI is trained over bacalhau, model access is that sold online with kickbacks to the data creators.
Total Budget: 5000
Budget File: pdf
Affiliations: none
LMIE Carveout: no
Team Skills: Blockchain developer, Fulltime, Economics bachelour, defi inventor, desci passion haver
Submission Number: 12
Loading