The customer is a Digital Media Advertising Technology Company that specializes in Real Time Bidding and Programmatic Media. Its recently launched Data Management Platform collects about 1.5 TB of adserving events, such as clicks, impressions, tracking data and so on from the web monthly per our Client’s Customers, who are more than a hundred.
The task was to analyze about 45 TB of raw data every day by means of more than a hundred rules written in human readable Domain Specific Language to build user-defined segments.
The solution was implemented from scratch by means of Apache Spark just in three weeks. It included Domain Specific Language (DSL) grammar, parsers, runtime representation of this DSL and the Spark Job itself. By the end of this period of time VIP Customers of our Client were provided with the ability to build its own online user segments and to perform advanced retargeting of the users of interest by means of human-readable language.