mikelev.in

From Jupyter Notebooks to Markdown: A journey through refining workflows, building muscle memory, and perfecting good form in data science.

mikelev.in

I’m currently deep in the process of integrating my professional and personal projects through Pipulate, focusing on how its three core components—the main framework, the specific workflows, and the Jupyter Notebooks used

mikelev.in

It begins with the satisfaction of a successful architectural improvement—converting Jupyter Notebooks for cleaner AI analysis—and then immediately pivots into an iterative debugging session.

mikelev.in

I was so close to a clean, elegant import pattern for my Jupyter Notebooks, just like the import numpy as np I was used to. But it kept breaking.

mikelev.in

While tools like Nix and Vim help bridge the gap, I hit a snag with Jupyter Notebooks polluting my Git history differently on each OS.

mikelev.in

It started with a nagging Git/Jupyter metadata problem (nbstripout) and evolved into designing a fully robust, user-friendly workflow.

mikelev.in

I am fighting to keep the individual maker relevant in an era of massive knowledge consolidation. By perfecting the Pipulate onboarding, I’ve developed a method to keep specialized context local, using a ‘Magic Wand’ that renders clean UI and clear speech simultaneously. It is a return to the autonomous workbench, where the user dictates the state, not the cloud vendor.

mikelev.in

This entry really captures the essence of how I’m approaching the AI-assisted refactoring process. It highlights the methodical steps: identify, distill, validate, and then integrate. The truncation incident was a great practical example of needing to ‘course correct’ the AI, and my ‘time-traveling prompt’ technique proved very effective. The ‘big but’ about DataFrame persistence is a crucial pragmatic decision that keeps the momentum going without sacrificing long-term goals. It’s a constant ba

mikelev.in

By anchoring generative intelligence within the linear constraints of Jupyter and Nix, I am proving that reliability is a choice.

mikelev.in

This technical journal entry details the iterative refactoring of a Nix Flake to automate Jupyter notebook setup, embracing a ‘chisel-strike’ development philosophy.