Seattle Apache Flink Meetup: Flink@Lyft and graph processing with Flink

Welcome to the Apache Flink community in Seattle! This time we are thrilled to have two speakers to talk about how they leverage Apache Flink for massive data stream processing:

• Gregory Fee, Principle Engineer at Lyft
• Theodore Vasiloudis, committer of Apache Flink and Machine Learning Researcher at Swedish Institute of Computer Science

Please share this with anyone who may be interested in Apache Flink and stream processing! We look forward to seeing you soon!

DATE: Jun 28, 2018 (Thursday)

LOCATION: 83 South King Street Suite 201, Seattle, WA 98104
(Please sign NDA at the gate to enter Lyft's office)

TALK #1: Making real-time decisions with Apache Flink at Lyft (~45min)

Speaker: Gregory Fee, Principle Engineer at Lyft

Bio: Gregory has been an engineer at Lyft for 2.5 years. Currently his work focuses on Dryft, a Streaming-Platform-As-A-Service project built using Flink. Dryft is the platform powering real-time machine learning feature generation at Lyft as well as real-time decision making and orchestration. Gregory has also worked on the Data Science Platform, Mapping Infrastructure, and ETA teams while at Lyft. Prior to Lyft, Gregory worked at Salesforce and Microsoft with a focus on programming language design and execution.

Abstract:
Lyft's mission is to provide the world's best transportation. To do this we need to make many coordinated decisions in real-time. Lyft has adopted Flink as the streaming technology to power the next generation of real-time decision engines.

This talk will cover some of the types of decisions that Lyft makes in real-time, the challenges faced with older solutions, and how Flink helps us better address those challenges. The talk will also give an overview of Dryft, the Streaming-Platform-As-A-Service solution at Lyft that reduces the complexity of building Flink applications and powers many of these decision engines.

TALK #2: Graph processing and cardinality approximation on massive streaming data with Apache Flink (~45 min)

Speaker: Theodore Vasiloudis, committer of Apache Flink and and Machine Learning Researcher at Swedish Institute of Computer Science

Bio: Theodore Vasiloudis is a committer of Apache Flink, focusing on Flink‘s machine learning library - FlinkML. He is currently a PhD candidate at KTH Royal Institute of Technology, and Machine Learning Researcher at Swedish Institute of Computer Science.

Abstract:
In this talk we will present some recent advances in analyzing and learning from streaming data from our lab at the Royal Institute of Technology and the Swedish Institute of Computer Science.

We'll present our work on Gelly Streaming, Apache Flink's graph processing library that allows analysts to derive in a single pass useful knowledge such as connected components and distance estimation. We will show how our most recent work enables iterative processing on top of streams, allowing algorithms like PageRank to be run on graph streams in a unified Flink pipeline.

We'll also demonstrate recent work that integrates Yahoo's DataSketches library into Flink, allowing resource-efficient approximation of cardinality statistics on massive streaming data.

Finally, if time permits, we'll showcase some recent work on parallel and distributed streaming learning algorithms, focusing on efficient distributed boosted trees.

AGENDA:

• 5:30pm - 6pm Food and networking
• 6pm Presentations and Q&A

EVENT SPONSOR this time: Lyft Seattle

Food and drinks will be provided.

TRANSPORTATION: Taking public transportations would be convenient. There're also street and inside-building parking spots.

**************
**************
If you are interested in giving talks and sponsoring our next event, please contact @Bowen.

We are looking for speakers for our August meetup.
**************
**************