Amazon SageMaker in Practice

project image
During the workshop, which is an adventure itself, we build an approximation of an ad targeting system with the use of Amazon SageMaker – from development to deployment. We show how you can start small with use of this service and be pretty confident about scaling up and expanding the system later.

Wojtek Gawroński
Principal Cloud Architect

Wojtek Gawroński

For whom?

  • People interested in learning more about AWS SageMaker and operational side of ML in the cloud.
  • People interested in learning more about AdTech and Ad Targeting.
  • People interested in learning more about Machine Learning and Data Science

Any requirements?

There is no need for previous experience with advertising systems and AWS either, although general knowledge about machine learning or data crunching are a plus.

What are the topics?

  1. Introducing AdTech and RTB domain. It will cover everything that you should know to understand the problem that we want to solve in this workshop. We will also address user targeting, GDPR and COPPA here.
  2. Introducing the problem that we want to solve - CTR (Click Through Rate). We will explain why it is important in real world, will describe the input data and will show how the machine learning can be helpful in this case.
  3. Introducing AWS and Amazon SageMaker. That will cover:
  • High-level overview of AWS
  • Introducing SageMaker and comparing it to the other machine learning tools
  • Explaining the life cycle of SageMaker application
  1. Applying XGBoost algorithm to solve the problem. We will first explain the algorithm and we will implement it using Jupyter.
  2. Deployment into the production environment. We will explain how it can be accessed from the outside. Also, we will address methods of scaling it up.
  3. Testing and monitoring. We will present and use AWS tools for that.
  4. Optimizing the algorithm using hyperparameters.
  5. Summarizing the costs.

Want to hear more?