Notebooks as Serverless Functions
summary
Jupyrest claims to turn Jupyter notebooks into HTTP functions by pairing a small config.json (ID, input/output JSON Schema) with a notebook that calls save_output and yields either a JSON result or a full notebook HTML; it runs as a serverless-style service (Azure Functions backend, optional execution backends like Synapse or Databricks), issues execution IDs, validates inputs (pydantic/JSON Schema + reusable Python model refs), ships from production use since 2020 and is now open source under MIT.
AI-written summary · about
chapters (10)
- 0:00Notebooks and serverless functions
- 3:59JupyRest and how it works
- 7:35Convert notebook to REST API
- 13:20Sharing models for scalability
- 20:42Runbooks and incident automation
- 23:53Deploy on Azure Functions
- 26:17Open source plans
- 27:00Next steps for project
- 28:07Deployment requirements
- 30:14Easy deployment button
related talks
24mRest Easy with Jupyrest: Deploy notebooks as web services
17mNotes on Notebooks: Is Jupyter the Bringer of Jollity?
26mAutomating runbooks with Jupyrest
34mNotebooker: Production and Scheduling for your Jupyter Notebooks
37mJupyterLab
25mOvercoming 5 Hurdles to Using Jupyter Notebooks for Data Science, by the JetBrai...
30mNotebooks in VS Code
24mJupyter Notebooks for Ops