[Machine Learning Platform Re-Evolution: End-to-End MLOps]From ML platform to ML ecosystem system, a full set of multi-person collaboration ML processes can be automatically generated without writing programs

As early as 2020, Line created a machine learning platform MLU (Machine Learning Universe) on the internal data platform IU (Information Universe), with the aim of hoping that MLU will become a data center just like IU plays the role of a single data entry point. The single ML entry for scientists or engineers can be used to schedule GPU resources, upload ML programs to train and test ML models, and can also be quickly deployed as external ML services. The goal of this MLU platform is to provide an MLOps platform that integrates the three workflows of ML, Dev and Ops.