A public good that would be really useful is a public dataset of embeddings for casts. 

Assuming it's legal to do, would drastically reduce the cost of building embeddings based models on Farcaster

Technowatermelon. Elder Millenial. Building Farcaster. 

nf.td/varun

I have build this partially with /amp (powering search and perplexity-like experiences) and have access to turbopuffer to be able to scale it, but not the resources. I would be interested in building an open source indexer and hosting an open API if there was a sensible way to fund it. Would you consider a Merkle grant?

building /amp and /guess. prev: 2x founder, improbable, github, amazon. ntik.me

/amp indexes already over 125k of the most popular casts.