Real-Time ML Applications at Scale with Flink and Beam on Kubernetes and GCP
summary
Rapido runs a production streaming platform that uses Flink on GKE to make near‑real‑time dispatch and ML scoring decisions, integrating Presto for analytics and MongoDB as a source. Infrastructure is provisioned with Terraform, CI/CD deploys Flink jobs, and Kubernetes Operators handle cloud‑native scaling and rolling updates. Observability relies on dashboards, logs and job metrics; the platform has run in production for two years and handles roughly 4,000 concurrent dispatch requests.
AI-written summary · about
chapters (8)
- 0:00Introduction to real-time ML
- 6:51Architecture overview and use case
- 13:06Kubernetes and GCP setup
- 20:23Flink and Beam integration
- 28:30Job management and scaling
- 35:03Demo: Real-time pipeline
- 48:16Handling failures and rollbacks
- 54:18Q&A and closing remarks
related talks
26mKubernetes & real-time analytics: How to connect these worlds with Apache Flink
30mCloudflow: Spark, Flink, and Akka Working together on Kubernetes
54mEconomics and Best Practices of Running AI/ML Workloads on Kubernetes
37mSpark and Flink Running Scalable in Kubernetes #bbuzz
35mEvolving and Supporting Stateful, Multi-Tenant Decisioning Applications in Production
37mIntegrate Flink with Kubernetes natively
35mBringing Your Data Pipeline into The Machine Learning Era
37mStateful Stream Processing with Apache Flink 1.5 and beyond