MLOps for a hybrid AI platform compliant with European law
summary
Describes a hybrid AI platform deployed on OpenStack with Kubernetes containers and extended to Google Cloud using Anthos, running edge inference over 5G (tested with a Boston Dynamics Spot Lidar pipeline) and central GPU training. Integrates a commercial DataIQ layer covering about 90% of the ML lifecycle alongside an open R&D stack using MLflow and Kubeflow for experiment tracking, CI/CD, model registry and drift monitoring, and argues these MLOps practices can meet EU AI Act requirements and support regulatory sandbox testing.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:58Birex Life Science context
- 5:49EU AI Act goals impact
- 10:14AI Act compliance challenges
- 14:36MLOps benefits
- 20:36Software vs AI in healthcare
- 23:23Regulatory sandbox
- 25:57Hybrid AI infrastructure
- 31:39MLOps tech choices Data IQ
- 39:46Integrating MLflow CubeFlow
- 43:20Future best practices
- 45:08MLOps for AI Act compliance
- 48:47Q&A
related talks
31mProductionizing ML with ML Ops and Cloud AI
49mKubernetes and MLOps for Scalable and Reproducible Generative AI
17mDevOps for Private AI: A Scalable Platform Using Kubeflow
26mKubernetes and MLOps for Scalable and Reproducible Generative AI
05mThe Evolution of MLOps
37mOptimizing AI and ML with Platform Engineering
29mTeam Teardown: MLOps in the Cloud for IoT
22mBuilding a MLOps Platform with Kubeflow on GKE