Building the platform for providing ML predictions based on real-time player activity

At Phoenix Games we are part of the mobile gaming industry and provide technology to our studios for delivering personalized content to the players based on real-time event processing and machine learning model predictions.

We developed our own infrastructure to ingest, process, and store vast amounts of in-game data and deliver customized recommendations in real time to improve the player’s experience.

Leveraging AWS Cloud technologies and Apache Flink enabled us to build an (almost) serverless mobile game event collection and processing infrastructure.

We want to share our journey from an engineering perspective and discuss our learnings from assessing various technologies.