Cool. It looks like you're packaging the data as json files in the repo. Would be neat to see the pipeline that collects it.