SnowCamp 2019

We will present a case study for the industrialization of a machine learning project with Apache Spark.
Starting from a prototype developed by a Data Scientist we will explore the unit testing problem and how BDD testing can help to build a easy to understand testing scenario.

We will then have a look at the scheduling and dependency problem and a possible solution with Azkaban.
To conclude we will talk about packaging, deployment and monitoring presenting the current solution used at Kelkoo Group.

We want to share our experience on best practices and tips when putting machine learning into production while using good coding and testing practices.