Built by Alex Kazos · Substack · GitHub. Inspired by Designed to Scale.
How do training and serving get the same feature, computed once?