Amazon has just launched SageMaker Studio, which it describes as the first IDE (integrated development environment) for machine learning.
This is another big move which happened over at the AWS re:Invent 2019 cloud conference, with SageMaker Studio aiming to be a one-stop-shop for developers and data scientists hammering out their machine learning workflows.
SageMaker Studio covers the whole process of building and training machine learning models, and then deploying and subsequently managing them – at any given scale.
Amazon notes that traditional machine learning development is an unnecessarily complex and convoluted process which involves trying to stitch together various tools, and all that palaver isn’t necessary with its new service, which delivers everything under one umbrella.
The company observes: “SageMaker Studio unifies at last all the tools needed for ML development. Developers can write code, track experiments, visualize data, and perform debugging and monitoring all within a single, integrated visual interface, which significantly boosts developer productivity.”
SageMaker Studio tracks all steps in the machine learning workflow, and makes it easy for the user to do things like clone them, or tweak them and then run them again, swiftly observing differences and generally making obtaining a suitably honed outcome a quicker process.
SageMaker Studio consists of a number of fully integrated tools to cover the full gamut of the machine learning workflow. They include SageMaker Experiments which allows for the tracking and organization of thousands of machine learning jobs (training, data processing, or model evaluation). And there’s a Debugger tool which automatically analyzes and troubleshoots training issues, providing real-time alerts on ways to optimize your model training.
SageMaker Model Monitor detects quality deviations in deployed models – again delivering appropriate alerts – and allows users to easily visualize potential issues such as data drift, all in a few clicks, Amazon explains.
And SageMaker Autopilot builds models automatically, as the name suggests, giving you full visibility into how it has made all the decisions.
Finally, SageMaker Notebooks is also now in preview, and delivers an “enhanced notebook experience” which lets users create and share Jupyter notebooks with a minimum of fuss – in fact in a single click. Amazon explains that the underlying compute resources are elastic, and easily scalable on-the-fly with no need to interrupt what you’re doing.