AWS SageMaker - Introduction

AWS Nov 3, 2021
"The purpose of software engineering is to control complexity, not to create it."

Typically there are a lot of moving pieces that are needed to be taken into account when we think of making our solution AI-enabled. Apart from business challenges, various technical questions need to be answered.

Here are a few challenges a Data Scientist faces:

  1. Managing various data sources securely.
  2. Annotating data.
  3. Experimenting with various training algorithms and analyzing the best suited.
  4. Hyperparameter tuning.
  5. Deploy the model on cloud or edge devices.

And a lot more. This list can go on.

But wait, is there any solution here?

Most of these challenges can be resolved using powerful, useful, developer-friendly, and feature-rich services offered by Amazon called SageMaker. The service was introduced by AWS in 2017 and is continuously evolving ever since.

What is SageMaker?

If you are working on AI domain, you would have heard something about AWS SageMaker.

SageMaker is an on-demand service offered by AWS which can take care of many tasks involved in the AI product development lifecycle. Fully managed machine learning service which takes care of data gathering, model building, training, deployment, and monitoring for a production-grade system.

Evolution of SageMaker over time

If we look back in time, it was not long ago when the service was launched by AWS in 2017 re: Invent conference. Ever since they haven't looked back for a moment. Let's visualize the journey SageMaker has gone through over the years:

2017: SageMaker is launched at AWS re: Invent conference

2018: TensorFlow and MXNet neural network support added within SageMaker. Support for Recurrent Neural Network, word2vec, and multi-class liner learning network added. Batch transformed introduced to perform non-realtime inferences. SageMaker Ground Truth service was introduced. It makes labelling of data much more simpler and fun. Reinforcement Learning was introduced with SageMaker, making it easier for developers to implement RL. SageMaker Neo was introduced to support edge computing.

2019: SageMaker Neo made open-source. SageMaker Studio launched. Feature to unify all the tools needed to develop ML models.

2020: SageMaker Feature store launched. A fully managed repository that helps maintain consistency between features used at the time of inference and model training. SageMaker Jumpstart launched.

Capabilities of SageMaker

SageMaker offers multiple features/services under its umbrella. These features assist developers in every stage of the AI journey. Let's discuss some of the services offered by SageMaker that stands out:

SageMaker helps developers at every step of the AI product lifecycle. Starting from data collection, to the deployment of production-grade systems.

Conclusion

We saw some of the most commonly used features from a very high point of view. In future articles, we will dive deeper into these services and see how can we make use of them and develop scalable solutions faster and with the least amount of effort.

See you in future articles, stay tuned for more detailed discussions related to Artificial Intelligence. Till then stay healthy. 😇

Tags

Arpit Jain

Machine Learning Engineer

Great! You've successfully subscribed.
Great! Next, complete checkout for full access.
Welcome back! You've successfully signed in.
Success! Your account is fully activated, you now have access to all content.