Summer is nearly here and it's time for our next Spark+AI Prague meetup! What can you expect this time? David Vrba from Socialbakers will be having a talk about UDFs in Spark SQL and Michael Shtelma from Databricks will talk about open-source project MLFlow. After the sessions, we will be having a BBQ on the terrace at Socialbakers HQ office!

==============================
17:30 - 18:00 - Welcome
18:00 - 18:45 - David Vrba (Socialbakers): UDFs in Spark SQL
18:45 - 19:15 - Break
19:15 - 20:00 - Michael Shtelma (Databricks): MLFlow in action
20:00 - 22:00 - BBQ and community building

The presentations will be delivered in English.

More about the talks:

David Vrba (Socialbakers): User Defined Functions in Spark SQL
In Spark SQL user defined functions (UDFs) are a technique how to execute a custom Scala, Java or Python code as a column transformation on a Spark DataFrame. It is widely understood that this flexibility is compensated by performance penalties. In this talk we will go over the possibilities that Spark SQL offers in regard to UDFs with Scala and Python and see what benefits were introduced by integrating PySpark with Pandas and Apache Arrow. Also we will discuss how the demand for UDFs was mitigated by releasing Higher Order Functions in Spark 2.4. This talk will be more theoretically oriented but we will also give some tips how to use these techniques in real-life queries and show some performance benchmarks to see how they differ in execution time.

Michael Shtelma (Databricks): MLflow in Action
MLflow is an open source platform for managing the end-to-end machine learning lifecycle. It allows experiment tracking, machine learning models packaging in a reusable, reproducible form, and managing and deploying them from a variety of machine learning libraries to a variety of model serving platforms. During this talk, you can learn, how to get started with MLflow, track the basic metrics of your models and model artifacts in MLflow, package your models as MLflow projects and how to deploy your models using MLflow in the cloud.