Talk Python To Me #530: anywidget: Jupyter Widgets made easy
79 snips
Dec 13, 2025 Trevor Manz, a software engineer and anywidget creator with a PhD from Harvard focused on data visualization, dives into revolutionizing Jupyter widgets. He discusses the challenges of integrating front-end UI with Python and how anywidget simplifies this by minimizing JavaScript usage. Trevor highlights its role in enabling interactive visualizations and data exploration. He also covers the importance of cross-platform functionality and using AI to streamline widget development, making advanced data interactions accessible for all Python users.
AI Snips
Chapters
Transcript
Episode notes
Altair Gains Bidirectional Selections
- Altair adopted AnyWidget to enable bidirectional selection that returns data frames into the kernel.
- That turned interactive visuals from isolated views into programmatically accessible inputs.
Drawing Data To Teach ML
- Vincent Womerdam built a Draw Data widget so users can sketch datasets and export points as data frames.
- Educators use it to visually teach how classifiers behave on different data shapes.
Widgets As Composable Notebook Blocks
- AnyWidget treats widgets as composable building blocks you can wire across notebook cells.
- In Marimo, updating a widget can automatically re-run dependent cells via the reactive graph.