End-to-End MLOps with MLflow and Kubeflow
summary
Uses MLflow to log Iris dataset runs with a logistic regression, logging hyperparameter arrays and an accuracy metric, creating an MLflow tracking UI instance, registering the model as registered model v1 (created version 1 after a 31s run) and loading it back as a Python function. Spins up Kubeflow on Kubernetes via CG devx, shows notebooks-as-pods, KServe for serving/preprocessing/explainability, training operators for TensorFlow/PyTorch/XGBoost, and compiles Python KFP pipelines into YAML that run and turn green in the UI.
AI-written summary · about
chapters (11)
- 0:00Introduction to MLOps
- 3:37Why MLOps matters
- 6:55MLflow core features
- 10:29MLflow tracking demo
- 15:44MLflow project packaging
- 17:37Kubeflow introduction
- 19:27Kubeflow notebooks
- 21:56Kubeflow pipelines
- 26:38MLflow vs Kubeflow
- 28:47MLflow Kubeflow integration
- 29:28Q&A
related talks
01h 25mIntroduction to MLOps with MLflow
31mMLflow: Platform for Complete Machine Learning Lifecycle
14mKubeflow: Where ML Meets Kubernetes
22mBuilding a MLOps Platform with Kubeflow on GKE
33mOperationalizing Model Lifecycle with MLFlow
20mML Ops design patterns with Kubeflow Pipelines
14mMLOps: DevOps for AI with KubeFlow Pipelines
30mMLflow: an Open Source Tool for Managing the ML Project Lifecycle