What are the MLOps tools?

Common tools are DVC for data versioning, ClearML, AWS Sage Maker, Neptune and Qwak for experiment management, Aporia for model monitoring. Some tools are expanding to include more functions in one tool as CoreAI (On Premise solution) and IguazIO. Machine Learning Operations is software versioning management on steroids combined with DevOps. Reconstructing an AI / ML experiment resulting with a model, holds a high number of variables involved, such as hyper parameters, versions of the data itself. Additionally, moving to production is more than CI/CD wrapping of the model and preparing it for serving, it's also the ability to monitor the performance of the model and detecting drifts in the data. Last, is the ability to bring the domain expert into the loop, and analyze the changes together with the Data Scientist. In summary, the main components of MLOPS are: Data Versioning | Research and Experiment Management | Production model performance | Research Process Automation | CI/CD (Model serving) | Data Monitoring