Episode 293: Agentic Data Science Pair Programming With marimo pair

The Real Python Podcast

May 01, 2026 1h 4m intermediate ai data-science

How do you add agent skills to your data science workflow? How can a coding agent assist with data wrangling and research? This week on the show, Trevor Manz from marimo joins us to discuss marimo pair.

Trevor is a founding engineer at marimo, where he’s been working on integrating LLM tools with marimo. We discuss the balancing act of building a skill and determining how to give an agent access to all the variables in a notebook. He shares how they built a specialized reactive REPL that eliminates hidden state and allows the agent to continue constructing a reproducible Python program.

We dig into installing and getting started with marimo pair. Trevor also covers several of the tasks an agent can tackle in a data science workflow.

Video Course Spotlight: Getting Started With marimo Notebooks

Discover how marimo notebook simplifies coding with reactive updates, UI elements, and sandboxing for safe, sharable notebooks.

Topics:

00:00:00 – Introduction
00:02:26 – Trevor’s role at marimo
00:03:08 – Current AI tools in marimo
00:06:26 – Describing marimo notebooks
00:10:11 – What is marimo pair?
00:18:49 – Building an agent skill
00:27:34 – Setup & installation
00:31:16 – Video Course Spotlight
00:32:42 – Examples of EDA and data wrangling
00:45:46 – Experimenting inside of a notebook
00:50:40 – Managing context
00:53:25 – Accessing additional libraries
00:57:16 – Recent tools and updates from the marimo community
00:59:31 – What are you excited about in the world of Python?
01:01:10 – What do you want to learn next?
01:02:26 – How can people follow your work online?
01:03:13 – Thanks and goodbye

Show Links:

Level Up Your Python Skills With These Courses:

Marimo: A Reactive, Reproducible Notebook

Getting Started With marimo Notebooks

Investigating Quasar Data With Polars and Interactive marimo Notebooks

Getting Started With Claude Code

← Previous All Episodes