The data scientists are gone. how do I deploy their work to prod?

on YouTube ↗40m

" summary

Shows how to convert leftover Jupyter notebooks, CSVs and slides into production-ready code by importing and refactoring notebooks inside VS Code, then packaging that work into repeatable pipelines. Demonstrates using Azure Machine Learning workspace (elastic compute, model registry, Docker images) for training, versioning and deployment, and using Azure Pipelines triggers and SDK calls to automate training, evaluation, packaging and conditional deploys so a new model replaces production only when metrics improve and workflows are testable and reproducible.

AI-written summary · about

" chapters (12)

1. 0:00Session presentation
2. 1:30The data scientists are gone
3. 4:00Going into production
4. 7:17Misunderstanding
5. 9:18The steps
6. 11:00Visual Studio Code
7. 24:40It's not perfect
8. 26:36Part 6
9. 27:17Part 7
10. 28:48Part 8
11. 29:30Part 9
12. 32:55Part 10

" related talks

• 25mFrom Data Exploration to Modeling to DeploymentRoope Astala · O’Reilly Jupyter Pop-up Boston 2018
• 45mMLOps and ML-Flops: going from dev to prodAndrew Blance · SQLBits 2023
• 57mDevOps for Machine LearningDamian Brady · NDC Oslo 2019
• 17mProductionizing Data Science WorkloadsAnand Iragavarapu · PyCon India 2020
• 37mPracticing Data Science in the cloudDr. Robert Hoffmann · WeAreDevelopers AI Congress Vienna 2018
• 19mFast Focus: Azure Notebooks for Artificial IntelligenceJennifer Stirrup · SQLBits 2020
• 43mA lap around Azure Machine Learning servicesPeter Drougge · Update Conference Prague 2018
• 39mSimplify, speed and improve development with DevOpsGerwald Oberleitner · WeAreDevelopers AI Congress Vienna 2018