Building a Complete, Real-time Timeline of User Activities at Netflix with Flink

Netflix strives to bring joy to its customers with personalized and relevant content. Proving implicit and explicit functions to understanding user behavior. These functions have contributed to building high-quality recommendation systems models.

Join our Meetup on May 23rd at 2pm PDT to learn:

•    How Flink Redis and Cassandra constructs a timeline to user activity in real time
•   A deep-dive into RockDB backed by Flink state and coupled with Redis as a data-serving layer
•   A discussion of several performance optimizations done to operate the system at Netflix scale
•   Challenges we faced around supporting as-fast-as-possible updates, while minimizing duplicate writes and dropped messages
•   A view on general scaling issues

Find out how Pinterest is utilizing Flink-powered stream processing to power its engine. Pinterest has onboarded 90+ applications that significantly transformed data processing and serving from batch to (near) real-time flavor in a range of use cases from Experiment Analytics, Metrics Aggregation, Ad Reporting, Shopping Catalog Ingestion, Rich Content Signal Generation and ML inference. Join our meetup to discover how Pinterest provides a buildable and deployable skeleton project (code / test / configs) with job developers plugging in their business logic with Ververica Flink. With CI/CD, load testing framework and job debugging tools, Pinterest is able to significantly reduce the time from prototype to production with sufficient job validation and minimal human efforts involved.