Artificial Intelligence (AI) and Machine Learning (ML) are two of the most popular technological terms in the twenty-first century. Although researchers first utilized them, many large corporations are now employing them to solve their business problems efficiently. Implementing machine learning models required a lot of physical computing resources with a lot of configurations before cloud computing, but now cloud providers like AWS, Microsoft Azure, and Google Cloud Platform supply a lot of virtual resources for ML modeling. Though they offer several managed low-code or no-code solutions for business problems, most scenarios necessitate custom training and tuning of machine learning models, which may require a lot of computing power.

AWS offers SageMaker services, which include various Machine Learning services for corporate applications, to meet the need for a flexible, hosted, and configurable ML development environment. Though SageMaker is not new to the cloud, AWS went one step further and created The SageMaker Studio, a Cloud IDE that houses all SageMaker’s services under one roof.

1. The Studio

AWS SageMaker Studio is a one-of-a-kind service providing multiple services under a single UI aiding the process of ML model building and fine-tuning of parameters. Firstly, to start a SageMaker Studio, we need to create a Domain. A SageMaker Domain consists of an Elastic File System (EFS) along with a list of authorized users, security features, applications, and VPC configurations that secure the environment. We can create a domain using either SSO Authentication or IAM roles.

2. The User Interface

Once the domain is ready, we can launch the studio environment for a particular user. There can be multiple users in a single environment. The SageMaker studio resembles an IDE running in the comfort of the cloud. It is very responsive, flexible, and functional, allowing us to alter the environment within a blink of an eye. The studio is equipped with a quick launcher that provides one-click solutions to launch a Jupyter notebook, python terminal, or prebuilt models using Jumpstart.

AWS Sagemaker Studio consists of 4 main parts:
a. Left Sidebar:

The left sidebar consists of several icons which allow us to switch quickly between the resources. It includes icons for:

File Browser: Displays the files and resources open in the main window.
Git tools: Provides Git integration to the studio.
Running terminals and Kernels: Displays the instances and kernels currently running in the session.
Commands: Provides controls for the studio and the notebooks running in the session.
Notebook tools: Provides the metadata about a notebooks instance.
Open tabs: The list of tabs open in the main window
Jumpstart: One-Click ML solutions, endpoints, and training jobs on various scenarios.
SageMaker components and registries: It is a customizable pane. Can be used to view the projects, jobs, pipelines, experiments, endpoints, data wrangler, etc.

b. Right SideBar:

The right sidebar consists of a settings pellet that dynamically shows us the resource settings that are currently on the main screen.

c. File and Resource Browser:

The File and Resource browser displays the list of open resources on the main working area. It also helps us launch new notebooks, upload files, create folders, etc.

d. Main Work Area:

The main work area displays the resource on a wide pane. It shows notebooks, terminals, features like data wrangler, feature store, etc., and allows us to work on it.

3. Studio Notebooks

SageMaker Studio revamped the SageMaker notebooks like the new Studio Notebooks. The studio notebooks are very similar to Jupyter notebooks with the additional configurations and look of the studio.

The above image is the UI of a SageMaker studio notebook. We can customize the studio notebook’s environment by changing the EC2 instance type, SageMaker Image, and the Kernel which runs the code. The upper toolbar in the notebook provides us control over it. Apart from the basic functionalities of a Jupyter notebook, in a Sagemaker notebook,

We can launch the terminal for execution.
Check the differences between current and previous commits in code.
Adjust the EC2 and Kernel configuration on the go.

AWS SageMaker studio notebooks provide all the EC2 instance types supported by SageMaker. It also provides us with a few fast-launch instances that launch faster than others and reduce the wait time. SageMaker also provides several images and kernels to select based on our requirements. We can also install our custom modules and packages before starting a notebook instance by setting a Launch-Configuration.

One of the most remarkable features of SageMaker studio is that it allows us to share our entire workspace and the environment configuration as a snapshot. We can share the snapshot with others or store it for future use.

4. Additional Studio Features

Apart from the notebooks, SageMaker studio integrates all other SageMaker features into its UI to use them with ease. The other features that can be used in a studio are,

SageMaker JumpStart: a one-click solution that accelerates ML workflows.
SageMaker Data Wrangler: seamlessly integrates and prepares the data using pipelines.
SageMaker Feature Store: a managed repository to store and retrieve ML features.
SageMaker Pipelines: an easy-to-use CI/CD service for ML.
SageMaker Autopilot: it is an Automated ML solution (AutoML) by AWS.
SageMaker Serverless Endpoints: used to deploy ML models and obtain inferences on the go.
SageMaker Model Registry: a central repository to catalog ML models.
SageMaker Projects: it helps you to create an end-to-end ML solution.
SageMaker Experiments: it helps you to track ML models.
SageMaker Inference Recommender jobs: help in choosing an instance for inference and
SageMaker Compilation Jobs helps to compile your ML models.

Conclusion

AWS SageMaker Studio is a one-of-a-kind service with a plethora of capabilities for building, deploying, and maintaining machine learning models. Though it is not for beginners, Data Scientists and Engineers working on real-time scenarios would find it helpful. In addition, features like Studio Notebooks, Quick Launcher, and Jumpstart will come in handy while developing an ML solution. SageMaker, as a whole, offers a diverse set of services to assist Data Scientists at every stage of the ML-Lifecycle.

I hope that this blog has given you a broader view of AWS SageMaker Studio. If you have any queries related to the same, you can drop in your questions in the below comment section, and I will get back to you quickly.

About CloudThat

CloudThat is Microsoft Gold Partner, AWS Advanced Consulting Partner, and a Google Cloud Partner and has successfully led many AWS-related implementation projects for our esteemed clients. So, get in touch with us for quick results.
Feel free to drop a comment or any queries that you have regarding cloud migration, and we will get back to you quickly. To get started, go through our Expert Advisory page and Managed Services Package that is CloudThat’s offerings.