August Presentation Night

Thank you to Databricks for sponsoring food and drink for tonight's event!

Agenda:

• 6 - 6:30: Food + Mingling
• 6:30 - 6:40: Opening remarks + Sponsor pitch
• 6:40 - 6:45: Lightning talk by Nicholas Chammas
• 6:45 - 7:20: Feature talk by Roope Astala

Lightning Talk: Robustly Packaging your PySpark Application on YARN
Speaker: Nicholas Chammas (MassMutual)
Intended audience: Running Spark in production
Abstract: Your PySpark application depends on some libraries from the Python ecosystem. How do you ensure your application can access these libraries without depending on a virtual environment or `site-packages` that is shared by multiple applications with potentially conflicting dependencies? I will show you how.

Feature Talk: Deep Learning and Computer Vision on Spark Made Easy
Speaker: Roope Astala (Microsoft)
Intended audience: Tinkering with Spark or running it in production
Abstract: I will give an introduction to our new open source library - Microsoft Machine Learning Library to Apache Spark - and discuss its architecture and design principles. I will present a walk-through of building a scalable computer vision pipeline that uses deep neural networks and transfer learning for image classification, and also discuss how to use a GPU VM together with a Spark cluster to train deep neural network models.