My summary of lessons learned applying data science.

Full writeup available here:.

Justfiles have been missing from my life and I only found out today. It's more flexible than `make` but the main killer features are that you can easily pass arguments to the commands *and* is that you can easily define a global justfile for all yer favorite utils!

RT @pipTrends: In this video, @fishnets88 & @GameMarketWest demoed some of the powerful features of Marimo Notebook. .

It sure is exciting to see a wave of new cloud providers that make it fun to quickly get started if you're a tiny team.

RT @jbhuang0604: Why More Researchers Should be Content Creators. Just trying something new! I recorded one of my recent talks, sharing wha….

I figured I should give this a spin and . yeah. Seems like there's a bunch of stuff to like.

Wrote a more detailed blog about how/why here:.

There are two tiers of open-source work in my life now. Tier one is good enough to put on pypi Iand tier two feels like it might be squatting names for personal tools. So I started a new habbit: the "installation" section of the readme.

Protip, if you want to blog more, build your own text editor. It's never been easier to make something bespoke and it totally works wonders.

The course also mentions mirascope, smartfunc and instructor. These are all great tools, but something about how llm just celebrates plugins makes it such an easy choice to get started with.

What tools does the course showcase?. Well . it's mainly llm from @simonw. Out of all the tools I've tried I found this one to be the most "boring in the best way possible". It's very straightforward, but still very extensible if you only know Python.

The reviews have been great sofar too, so if you want to learn more, you can check it out here:.

I made a new course with the folks over at @TalkPython about LLM building blocks. The material contains nothing wild. Just calm, simple methodical techniques that actually work within 5 minutes.

The prompt was "starcraft marine riding a vulture". Two different systems, two different interpretations and very hard to say which one is "better".

Python notebooks are the bees knees!.

If you're interested in a larger talk/demo about this work, you'll want to see our conversation from yesterday. Some good nuggets in this one. @HamelHusain thanks for hosting the talk!.

I'll put it bluntly: I made this so that YOU (yes YOU!) no longer have an excuse not to look at your data. You can now go to your employer and actually convince them that you need to expense a gamepad for work. It could not be simpler/easier/more fun to annotate now. So do it!.

The widget is super nice and if you want to give it a spin, you can check the Github Pages on the repo to get an experience right away. There's a notebook that runs in WASM so you won't need to install anything.

And then the widget takes care of the rest.

The examples and render function are needed to populate the HTML in the center. This needs to convery everything the user nees to assign a label. The user can also add a note besides the label, and you can use your voice to add this note.