Serverless, automated, personalized book recommendations at scale on AWS

Description

PRH is three years into an effort to build and make use of the book publishing industry’s largest and most valuable direct-to-consumer book marketing platform, and one aspect of it is the Momentum Engine. The Momentum Engine is a serverless application built on AWS using Lambda, DynamoDB, Kinesis, SQS, EMR and data pipelines, and S3. The Momentum Engine allows PRH to use continually updated, data-driven analytical models to automatically compose and send personalized book recommendations emails to its subscriber base—at scale.

Brian Young details PRH’s journey in engineering and building this system. He explores the overall context of the project, the vision and genesis behind it, and how PRH went about mingling its artists and soldiers—operationalizing innovation at PRH. You’ll see how successful PRH has been and its current system volume and reader engagement numbers. Brian describes the current stable architecture and aspects of the engineering journey that led it there.

He takes a deep dive into automated test suites, cost transparency, dev versus ops, and learning culture and leadership culture. A good automated end-to-end test suite enables you to refactor your way out of a tight spot into your target performance range and forward to a sustainable maintenance future; cost transparency helps engineers reduce operational cost and system complexity; the way PRH’s old dev versus ops tensions want to migrate to the cloud and how the company broke them down (and didn’t); and how PRH developed a team culture of learning, fostered by its leaders.