if designed carefully i think we can attack one fr...
# datascience
k
if designed carefully i think we can attack one front and leave the other open to the future. so in golem I designed it to support pluggable backends which wrap implementations on specific platforms, and then golem-core exposes the api. if you write it that way, you could wrap it in e.g. spark or some ML lib on jvm for now, and leave it open for future support on the other platforms