Ray Will Dominate

My conviction for a product has never been so high for any ML Ops framework as for Ray. Let's be honest, most "ML Ops" libraries suck, not just suck they will probably slow down your ML scientists and data scientists vs not even using anything. Occasionally (surprisingly quite often now), someone asks me about ray and why I think is going to win. I spent plenty of hours trying to formalize my thoughts and here is a summary of it. In layman's terms, through a set of beautifully designed libraries and easy-to-use decorators (@ray.remote),