Streamlining your Machine Learning projects with Azure Machine Learning
summary
Shows Azure Machine Learning running end-to-end MLOps: notebooks pushed to an Azure ML workspace, experiments tracked with exact script and parameter logging, hyperparameter sweeps scheduled on cloud compute (demo spins a cluster to six nodes and trains on a 16,000‑image dataset), best model put in a model registry and published as deployment as Docker image endpoints, hooked to telemetry for automated retraining, and integrated with Azure DevOps. Demos prove the pipeline works and that cloud compute is billed by runtime.
AI-written summary · about
chapters (22)
- 0:00Introduction to talk
- 2:13Azure ML for non-experts
- 5:22ML workflow in Azure
- 7:13Compute resources in Azure ML
- 10:43Building datasets and notebooks
- 12:10Hello world ML example
- 13:33Evaluating model performance
- 17:09Open-source tools and integration
- 18:14Iterating on experiments
- 20:23Scheduling cloud training
- 24:30Demo starting Azure ML
- 26:14Notebooks in Azure ML
- 28:44Running first experiment
- 31:34Handling false positives
- 33:18Cloud vs local training
- 35:03Scheduling training jobs
- 37:10Automating ML pipelines
- 40:22Exposing models as endpoints
- 41:47MLOps with Azure DevOps
- 44:16Monitoring and validation
- 46:01Collaboration in ML projects
- 46:23Q&A and closing
related talks
42mPython SDK to train, package and deploy models
43mA lap around Azure Machine Learning services
10mGet Started with Azure Machine Learning
44mPractical deep dive into machine learning techniques and MLOps
27mBeginners guide to Azure MLOps
21mMLOps for R with Azure Machine Learning
45mMLOps and ML-Flops: going from dev to prod
57mDevOps for Machine Learning