SPONSOR WORKSHOP / Mike, Joe, Ryan Soley, Sri / Capital One
summary
Open source rubicon-ml provides a Python logging library plus a dashboard and publishing workflow that captures hyperparameters metrics dataframes and artifacts to local storage or S3 and enforces git integration so each experiment records the git commit hash linked to GitHub for reproducibility. A scikit‑learn Pipeline example produces 120 experiments automatically and the Prefect + Dask example shows concurrent logging. Publishing emits a YAML catalog and publishing writes a YAML catalog not moving data.
AI-written summary · about
chapters (8)
- 0:00Introduction to Rubicon
- 1:38Core features and benefits
- 4:08Glossary and installation
- 5:55Quick start demonstration
- 14:20Git and version control
- 17:22Scikit-learn integration
- 22:18Prefect workflow integration
- 28:30Community and conclusion
related talks
17mR & Python playing nice, in Production
49mKedro + MLflow - Reproducible and Versioned data pipelines at scale
24mUncovering Community and Project Insights Through Data Driven Methods
33mBuilding Reproducible ML Processes with an Open Source Stack
22mImprove Your ML Projects: Embrace Reproducibility and Production...
33mCompliant Data Management and Machine Learning on Kubernetes
45mWorkshop: Deploy and Monitor ML Pipelines with Open Source and Free Applications
35mOrganizing data engineering around Kubeflow