Scoring Machine Learning Models at Scale [Video]

At Strata+Hadoop World, MemSQL Software Engineer John Bowler shared two ways of making production data pipelines in MemSQL:

1. Using Spark for general purpose computation.

2. Through a transform defined in MemSQL pipeline for general purpose computation.

In the video below, John runs a live demonstration of MemSQL and Apache Spark for entity resolution and fraud detection across a dataset composed of a hundred thousand employees and fifty million customers. John uses MemSQL and writes a Spark job along with an open source entity resolution library called Duke to sort through and score combinations of customer and employee data.

MemSQL makes this possible by reducing network overhead through the MemSQL Spark Connector along with native geospatial capabilities. John finds the top 10 million flagged customer and employee pairs across 5 trillion possible combinations in only three minutes. Finally, John uses MemSQL Pipelines and TensorFlow to write a machine learning Python script that accurately identifies thousands of handwritten numbers after training the model in seconds.

About the speaker: John Bowler is a Software Engineer at MemSQL. John has a background in machine learning, algorithms, and distributed data warehouses. John is a graduate of MIT who previously interned at SpaceX where he helped write control algorithms for the SuperDraco rocket engine.