How BMW stays on the fast track by building the Connected AI Platform for digital products
summary
Connected AI Platform goes live last year and powers BMW Connected Vehicle ML by combining Kubeflow on EKS with AWS managed services (S3, Kinesis, EMR on EKS, SageMaker, Glue). It moves DevOps off data scientists via cloud IDE notebooks, reusable Kubeflow pipeline steps, the KFP SDK, KServe model endpoints, Helm charts for Streamlit, and CI/CD using self‑hosted GitHub runners plus Terraform/Terragrunt. Multi‑region, multi‑env clusters (examples: Frankfurt, Virginia, Korea) enforce data isolation and run production use cases successfully.
AI-written summary · about
chapters (7)
- 0:00Connected AI Platform intro
- 1:32BMW’s AI strategy
- 5:49Platform features for data scientists
- 16:57Architecture and tech stack
- 22:35Product management for platforms
- 26:27Onboarding with use cases
- 33:14Lessons learned
related talks
27mData-driven and cloud-native ecosystems in the automotive industry
45mBuilding a connected vehicle and app experience
15mSheer driving pleasure: Automated driving for BMW’s Neue Klasse
11mAWS On Air ft. BMW & BCG
45mBMW Software Factory - winning strategies to maximize value
44mEngaging the Kubeflow Community: Building an Enterprise-Ready AI/ML Platform
14mKubeflow: Where ML Meets Kubernetes
12mLeveraging AWS and Kubernetes for Scalable AI/ML Workloads in the Cloud