Spark Streaming is based on the in-memory, data processing engine Spark. Rather than running iterative algorithms on top of a fixed data, Spark Streaming uses the same data abstraction as its batch processing engine, but applies it to a stream of data. At Rockerbox, Spark Streaming allows us to run custom iterative, analyses on streaming windows of data to update targeting and bidding strategies in real-time. In this talk, I will discuss how we use a static dataset in Spark to explore a new strategy , move this analysis to a streaming job and ultimately produce a quantifiable result that can be used to improve our buying.
Pizza & Beer will be provided.
Rick O'Toole : Rick O'Toole is the founder and CTO of Rockerbox, the leader in Momentum-Marketing. Rockerbox processes and analyzes over 4 billion ad impressions a day to determine the real-time intent of millions of web browsers. With this insight, Rockerbox enables companies like Shutterstock, Journelle and Learnvest to target the optimal user with the correct message at the right moment. Prior to Rockerbox, Rick led integrations at AppNexus for companies such as Facebook and Yahoo in addition to being a founding data science engineer in AppNexus’ Marketplace-Analytics group. Rick holds a B.S. in Mathematics and Economics from Harvey Mudd College.