I used straight python to hold my evaluation data structures, instead of putting them in JSON. I apologize for nothing.

Thanks Max! In later stages that's where I hope to find wins. A small model might be trained on your texts in an ongoing manner and accept some direct user feedback. (probably not a full-blown transformer, but other more amenable model types. this will be the major focus of future installments)

Examples of spelling error detection true positives, false positives, and false negatives.

The latest stage in my quest to build a small/local/purposeful artisanal language model: Giving it a job and writing evals

brandonrohrer.at/alms_task.html

ok so my ex yc vp is vv go go on ai rn bc he is in an sv vc gc or we -- my em is in on it w/ ai as an os to do ui qa in ci -- so tl dr ig im tl of ai ui qa ?? rn ai ui qa v1 is cc in an hv vm on my pc on gh pr xd

“AI can do the job of all managers, middle managers, VPs. But not mine. As the Big Manager I bring something unique, a special human-only contribution which is…ummm…uh…hold on just a sec I have something for this. “

Code review for data (and non-SWE) folks TL;DR is, your job as a reviewer is to help the reviewee come up with better code than they initially did.

This post by @randyau.com is great www.counting-stuff.com/code-review-...

"Reading other people's code is like peeking into their minds. It's weird. It's uncomfortable. It's hard to understand because nothing is where I expect things to be."

now it has a smaller unfriendly wall of links

Tutorials, projects, code, and thoughts collected into topic groups I've generously called Book Projects. Highlights New Releases Being a Staff+ Data Scientist in 2026 Build a custom tokenizer Artisanal Language Models Most Visited Transformers from scratch Setting up an ssh server How to convert RGB color images to grayscale Convolution in one dimension Most Loved Choose your professional path What to do when a leader does something wrong On microsuffering I'm most proud of Solving an easy reinforcement learning problem on hard mode: Inverting a pendulum Naive Cartographer: A Markov Decision Process Learner Ziptie: Learning Useful Features

My blog had evolved into an unfriendly wall of links. To fix this I added an on-ramp section at the beginning.
brandonrohrer.org

10/10, no notes. The one thing that I am curious about, though, is the potential that Bayesian methods have for solving or mitigating a lot of these issues. A lot of business problems do, after all, feature strong informative priors

I'm not a data scientist, but I recognize so many of the interesting challenges and quirks of being in a similar space from this fantastic article.

Being a Staff+ Data Scientist in 2026 Brandon Rohrer | became a data scientist in 2013 when the title was young. It was so new that most companies had no idea what a data scientist should be doing, only that they desperately needed one or they would be left behind. Sound familiar? I've tried to survey the job description of data science a couple of times with varying degrees of success, most recently to go with some informal recommendations for creating data science degree programs. Together with a group of colleages we tried to summarize what data scientists do and the data science subtypes of maker, oracle, detective, generalist. But in the face of changing expectations this doesn't feel like enough anymore. It's time for a refresh. A brief and biased history of the Data Scientist role In the beginning... The field of data science was named in 1997, and the discipline has existed by other names for a very long time. After all, people have been answering questions using data for thousands of years. When data science first got huge, organizations expected data scientists to spin straw into gold—to transform unorganized data archives into profit. Big Data, it was believed, held inherent value, which only needed to be coaxed into cash form. This rarely panned out, so the approach evolved into a) data scientists produce "insights" and then b) "insights" generate profit. This also proved elusive in the end.

Hey #datascience people, new blog in which I aim to describe the complexities of being a Staff+ data science (and adjacent) roles. Let me know what I missed.

brandonrohrer.org/ds_roles

It's not just you--job hunting today really is a beast. As @shappy.bsky.social and @johnathan.bsky.social put it, the front door is fucked.

The secret? Focus on connecting with actual human people.

www.rawsignal.ca/newsletter-a...

thanks for listening, you can resume your apocalypse watch now

I heard they migrated to mastodon

my inner 8 year old is giddy with the fact that I can just download hi-res images of THE MOON FROM SPACE and have them greet me from my monitor each morning. It's like I work in a rocketship.
images.nasa.gov

On AI, margin pressure, and outsourcing all the best parts of work.

Oh, and on inventing tomorrow.

How's that for clickbait?

A small white dog in a blue coat, standing in front of a towel, covered with lines of treats. Apples, popcorn, carrots, gingerbread cookies, and various others.

when you want to do lines but you're a shih tzu

YES! Where can I buy tickets for the sequel?
🍿🍿🍿

That's my secret Cap. I'm always righteously indignant.

Banger post this week!

What AI Is Actually Doing to the Workforce YouTube video by The Atlantic

This is what it sounds like when grown-ups talk about AI. Johnathan and Melissa are absolute legends.
(this is a @rawsignal.ca stan account)
youtu.be/ncmuXQGGqBM?...

The era of unscientific management It's a really annoying turn of events but here we are...

This week on Counting Stuff, we're now in an era of 'unscientific management' and the systems we have are most definitely not designed to all the grifters out at the speed of their yes-bot enabled bullshit #dataBS

www.counting-stuff.com/the-era-of-u...

although this recent version illustrated by @jemimacatlin.bsky.social is in a similar vein, and currently the object of my affection.

12 copies of J.R.R. Tolkien's The Hobbit, each in a different language.

I would love to find a copy of this. I'm not a collector by nature, but this is my favorite story. Of my collection, only the Russian has a level of vintage whimsy that compares.

where it's at

Hilary's Closet Hilary is using Poshmark to sell items from their closet. Poshmark is a fun and simple way to buy and sell fashion. Shop from millions of people—and start selling too!

I've been having a ton of fun in the jewelry / resale space (I mentioned this on the last episode of NSSD). I've been hesitant to share bc fundamentally I still get embarrassed that I'm a GIRLY GIRL. But here's my lil' shop: poshmark.com/closet/curat...

you will certainly not regret buying ten yards of stonewashed denim

The science of Artemis II YouTube video by nature video

I am biased but this is the greatest and possibly only Artemis II video you need 🌝 www.youtube.com/shorts/Tc2vN...

A watercolor and ink illustration showing the breakdown of king size bed space when shared by a couple (combined for 50%), their cat (6%), a dog (40%), and a mandatory heat buffer between them (4%). Text label pointing to the dog reads "40% Teddy, the magically expanding dog."

Day 1: Part-to-Whole

Visualizing Teddy's magical expansion from a very average 45 pound dog into an immovable dire wolf once he hits the bed. Watercolor & ink.

#30DayChartChallenge

hundreds of seconds now