Evolving with Kubernetes: Embracing Model Ops
summary
Claims Kubernetes can host the full ML lifecycle — data acquisition, cleansing, model development, packaging into images and deployment — and does so using container image registries, triggers for redeploy, GPU and accelerator scheduling, and runtime monitoring for model drift. Examples include production diagnostics in healthcare and large-scale anomaly detection in finance running on the OpenShift platform. The talk shows model ops pipelines that run as a continuous pipeline with auditability, portability, and autoscaling.
AI-written summary · about
chapters (7)
- 0:00Intro
- 0:13Data AI workloads
- 0:48Challenges
- 2:05Operation
- 2:45Agility
- 4:38Model Ops Pipeline
- 5:47Conclusion
related talks
10mOptimizing Machine Learning (ML Ops) in Financial Services with Kubernetes and Cloud Technologies.
24mDemo of Kubernetes at the Edge for AI Model Deployments
25mWhen Kubernetes Meets MLflow: Orchestrating MLOps Processes the Cloud Native Way
26mMachine Learning Meets K8s
39m3+1 Ways to Do MLOps with Kubernetes
26mCloud Native Machine Learning Systems at Day Two and Beyond
21mMastering the DevOps Machine Learning Pipeline
44mDevOps for Machine Learning - why is it different?