09-26, 10:05–10:55 (Europe/Amsterdam), Apollo
Python notebooks revolutionised our field, don't let those influencers on LinkedIn tell you otherwise!
Their power for rapid prototyping granted us so much creative freedom and dramatically lowered the entry barrier for newcomers. Yet despite all this ... maybe it's time to reimagine and rethink notebooks in Python. There's small things that could be better like git integration and collaboration ... but if we really allow ourselves to rethink a few things, we may just end up with a notebook that really makes us better at solving data problems.
This presentation will be a stream of demos that help paint a picture of what the future might hold. I will share my latest work in the anywidget/marimo ecosystem as I try to convince you that if you want better notebooks, reactive Python might very well be the future.
There are going to be many demos in this talk that will highlight what happens if we force Python to behave reactively instead of imperatively.
- Some of these demos will show notebooks widgets that help you play with your data by merging javascript/Python interaction.
- Others will focus on how notebooks can be made self reproducible by supporting UV/pytest directly in the notebook itself.
- But we will also explore how LLMs fit into the bigger picture, especially when you're interested to work with data. Want to zero-shot an interactive notebook around your data?
- Oh, and another fun thing, thanks to WASM you technically won't even need a local Python environment anymore for any of the things that I am talking about. I will demonstrate that too!