Introduction to Amazon SageMaker Studio

What Is Amazon SageMaker Studio?

Amazon SageMaker Studio is a web-based IDE built for machine learning. It brings together notebooks, experiments, training jobs, model deployment, and pipelines in one place.

If you have an existing ML project built locally or on a server, SageMaker Studio lets you scale that work without rebuilding it from scratch.

Why Migrate to SageMaker Studio?

Running ML projects locally has real limits. You hit memory walls, training takes too long, and sharing work is hard.

SageMaker Studio solves these problems:

• Scalable compute: Spin up GPUs or large CPU instances on demand. Pay only when you use them.
• Managed infrastructure: No servers to set up or maintain.
• Experiment tracking: Every training run is logged automatically.
• Pipeline automation: Automate your full workflow from data prep to deployment.
• Team collaboration: Multiple users can share a Studio domain.

The SageMaker Studio Interface

When you open Studio, you see a JupyterLab-style layout with extra panels on the left:

• File browser: Your project files stored in EFS (Elastic File System).
• Running terminals and kernels: Active sessions you can stop to save cost.
• SageMaker resources panel: Experiments, pipelines, endpoints, models.

Everything you do in Studio (training jobs, pipelines, endpoints) shows up in this panel.

Key Concepts Before You Start

Understanding these terms will help you migrate more smoothly.

Domain: A Studio domain is a shared workspace for a team. It connects to an S3 bucket, VPC, and IAM role.

User profile: Each person in a domain gets their own profile with isolated home storage.

Kernel gateway: The compute attached to your notebook. You can choose the instance type (e.g., ml.t3.medium for light work, ml.g4dn.xlarge for GPU).

SageMaker execution role: An IAM role that gives Studio permission to access S3, training instances, and other AWS services.

What Migration Looks Like

Migrating your ML project to SageMaker Studio is not a single step. It happens in stages:

1. Move your notebooks and code into Studio.
2. Update data access to read from S3.
3. Move training to SageMaker Training jobs.
4. Register your models in the Model Registry.
5. Deploy with SageMaker endpoints.
6. Automate with SageMaker Pipelines.

You do not need to do all of this at once. Many teams start by just moving notebooks into Studio and running training jobs, then add pipelines later.

Next Steps

In the next tutorial, you will set up your SageMaker Studio environment and get ready to start moving your project.