Dating with Models – LA High Scalability Group

June 11th, 2013, 0 Comments

For those of you in the Los Angeles area, we’d love for you to join us and the Los Angeles High Scalability Group at eHarmony while we discuss highly scalable development practices in eHarmony’s match-making models.

Date: Tuesday, June 18th

Time: 7pm – 9pm

Overview:  eHarmony was one of the earliest, fastest growing and still one of the largest online dating sites. From the beginning, eHarmony has had data centric models at its very core. Every pair of users that are introduced to each other must pass through psychological compatibility models. Years of matching and communication data is combed through to generate models that predict communication. Over the years eHarmony has evolved its models to more and more sophisticated models that require an ever increasing amount of data, both during training and even more critically during the match creation process.

This presentation, by eHarmony architect Ryan Barker, will start with a brief overview of of the various architectures spanning the last 13 years, followed by an in depth analysis of the current hybrid SOA + Hadoop based architecture. Machine learned dynamic code generation scala models, JMS and REST based services, spring wired Hadoop jobs and various back end processes all producing and consuming a protocol buffer based data model.


Food and Beverages will be Provided by MACRO.CCS.

(Psst: non alcoholic beverages only, though there are talks of doing happy hour afterwards at a Santa Monica bar if anyone is down…)





Share This: