I trained a model. What is next?

Here at Kaggle we’re excited to showcase the work of our Grandmasters. This post was written by Vladimir Iglovikov, and is filled with advice that he wishes someone had shared when he was active on Kaggle. The original post can be found on Vlad’s Ternaus Blog . Introduction I participated in machine learning (ML) competitions at Kaggle and other platforms to build machine learning muscles. I was 19th in the global rating , got Kaggle Grandmaster title. Every ML challenge ended with new knowledge, code, and model weights. I loved new learnings but ignored the value that old ML pipelines could bring. Code stayed in private GitHub repositories. Weights were scattered all over the hard drive. In the end, all of them were deleted. The situation is not unique to Kaggle. Same story in academia. T