DESCRIPTION

At Amazon Advertising, we are developing state-of-the-art large-scale advertising and real-time feedback based control application. The Budgeting and Pacing team develops systems that control Ad spend in real-time to adhere to the goals and constraints set by advertisers. Our systems process billions of ad impressions daily from across the internet to power our display advertising algorithms. Our engineers work with machine learning scientists, economists, and product managers on high impact initiatives for Amazon’s Display Advertising.

As an engineer on the team you will design and develop products that are at the heart of advertising – products that determine how Ad opportunities are purchased by advertisers and how much advertisers pay for those opportunities. You will have responsibility to help define requirements, create software designs, implement code to these specifications, define continuous integration testing and support products while deployed and used by our customers. You will have complete ownership of technology choices, architecture, A/B testing for your features, code deployments into production environment and operations for the products you own. Your team will include Product Managers and other high performing engineers. We’re looking for SDEs interested in developing real-time bid and optimization algorithms on large scale systems that process tens of billions of requests/ day.

From the center of our technology platform, you have the ability to work on some of our largest-scale and most technically complex problems and your contributions will be magnified across a variety of use cases.


About the team
The Control System contains a series of inter-connected optimization and feedback control systems, including Mentat (ADSP’s pacing engine) that aim to help advertisers maximize order and ad-level budget delivery and performance against their campaign goals. In this document, we give an overview of the current Control System and its system architecture, summarize its limitations, and propose a north star vision to address the Control System’s limitations.