Managing Machine Learning Lifecycle with MLflow
summary
Walks through MLflow setup, shows tracking server (SQLite backend + artifact store), records experiments from PyTorch training on Kuzushiji‑MNIST and logs metrics producing 86% accuracy, demonstrates saving models as MLmodel artifacts and a model registry with versioning and stages, packages projects with Conda or Docker, serves a model via local REST and shows deployment options (SageMaker, Azure, Kubernetes), and outlines reproducibility pitfalls and CI/CD integration.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 2:36Signify’s ML R&D
- 7:36ML lifecycle challenges
- 9:29MLflow components
- 13:12Tracking experiments
- 19:40Neural network tracking
- 28:26Model registry workflow
- 36:39MLflow use cases
- 39:44MLflow Projects
- 46:01Running MLflow projects
- 49:23MLflow Models deployment
- 55:44Q&A
related talks
31mMLflow: Platform for Complete Machine Learning Lifecycle
33mOperationalizing Model Lifecycle with MLFlow
45mMachine Learning Lifecycle Made Easy with MLflow
55mOrganizing the Machine Learning Lifecycle with MLflow
37mPlatform for Complete Machine Learning Lifecycle
30mMLflow: an Open Source Tool for Managing the ML Project Lifecycle
01h 25mIntroduction to MLOps with MLflow
43mMachine Learning Engineering principles with Python and MLFlow