Fast Data at ING

on YouTube ↗46m

" summary

ING builds a production streaming data platform using Apache Flink and Kafka that delivers actionable notifications and fraud detection. The platform processes roughly 0.5M events/sec with latency under 50ms in tests and meets operational targets of subsecond decisions for transactions. Flink CEP, sliding windows and online scoring run PMML models inside Flink; models are produced offline today and loaded at runtime via a Kafka control stream using a simple PMML models interoperability approach. Continuous retraining is planned.

AI-written summary · about

" chapters (7)

1. 0:00Introduction and background
2. 10:16Building the streaming platform
3. 22:54Fraud detection with Flink
4. 26:04Actionable financial insights
5. 31:27Why Flink at ING
6. 38:14Future ambitions and summary
7. 40:01Q&A session

" related talks

• 43mDevOps at ING AnalyticsG. D'alessio, T. Bakker · Codemotion Amsterdam 2018
• 25mData: Moving with the flowAlex Walden · Big Data LDN 2022
• 51mReal Time Investment Alerts using Apache Kafka & Spring Kafka at ING BankMarcos Maia, Tim van Baarsen · Spring I/O 2019
• 45mApache Flink — Real-Time Data ProcessingGaël Renoux · Devoxx France 2024
• 41mBeyond the Hype: When Apache Flink Solves Real ProblemsNaci Simsek · Berlin Buzzwords 2026
• 27mArchitecting Sub-Second Analytics PipelinesSai Kaushik Ponnekanti · Conf42 Machine Learning 2025
• 50mData Science in the EnterpriseBas Geerdink, Natalino Busa · J-Fall 2015
• 57mEmergence of Real-Time Analytics: Real-time Analysis of...S. Palthepu · Philly ETE 2016