Categories on Game B.

Jev and an LLM try to roll a fair die

Thu, 17 Sep 2026 22:40:28 +0200

Generated using ChatGPT Images 2.5.

There's been a lot of hype and discussion alongside TypeSafe's new model, Jev.

Jev is interesting:

An intuition for abstraction

Mon, 15 Dec 2025 22:40:28 +0200

Building an intuition around abstraction in code can take a little bit of practice and experience to learn. Let's make it easier.

In South Africa, you need to renew your driver’s license card every 5 years. This is a rite of passage – it involves forms, long queues, confusion, chaos, and eventually a new license card.

Here’s how it works:

You queue outside of the entrance.
At the front door there’s a person (usually a security guard that’s been there for years, and learned the pattern) who tells you which form to fill out, and which queue to stand in.
While waiting in the queue, you fill out your form.
Once it’s your turn, you walk into a room, hand in your form and have your eyes tested by a machine being handled by an eye test person.I have no clue what to call this person that handles the eye test machine.
You then get placed in a different queue by a clerk, hand in your form and your eye exam results to the teller, pay, and leave.
A few weeks later, you get an SMS, and go collect your new driver’s license card.

Building intuition

Let’s notice a few things:

Working with LLMs

Sat, 08 Feb 2025 22:40:28 +0200

When you're an LLM engineer building a product with LLMs in an org with other engineers who don't, you get a lot of the following on Slack:

User: “I tried X and it didn’t work.”

You: *investigates*

You: “Nothing looks obviously wrong, it might’ve just been bad luck with the completion. Does it happen all the time? What happens if you re-run the query?”

User: “It worked the second time I tried it.”

AI Friends

Tue, 24 Dec 2024 22:40:28 +0200

WHILE I write this, everyone is shocked by the death of a teen who was an obsessive user of Character.ai (bear in mind -- this post was largely written in October, when the news happened, before life got in the way). While details are still emerging, I think it's time to put to pen something that's been on my mind for a while now regarding the effect AI companions will have.

I’ve been thinking about the repercussions of generative AI on our social fabric for quite some time now. And the stage seems to be set, and the potential issues are only now entering the public eye.

What makes an A player?

Fri, 23 Aug 2024 22:40:28 +0200

AT OpenBB, we've been thinking about the attributes of high quality employees over the last few weeks.

Talent density at OpenBB is particularly high.The highest I’ve personally experienced in my professional career so far, in fact.This has largely come down to OpenBB’s philosophy of hiring slowly and firing fast when it becomes obvious someone isn’t a good fit. Because of how OpenBB functions, I generally agree with this approach.OpenBB gives employees and extraordinary amount of flexibility, freedom, and autonomy, but this is paired with high expectations (rightfully so). And we’re currently rethinking what “good” performance looks like.

Life update

Fri, 09 Aug 2024 22:40:28 +0200

IT'S been a long, long time since I left Praelexis, and a lot has happened since then.

You can read my goodbye email I wrote here.

It’s been a very busy period in my life, with ups and downs.

I’ve wanted to write about some of the things that have happened for a while now. But somehow, as the list of things I’ve wanted to write about grew, it somehow became more difficult to start. Kind of like what happens if you try to pour too much cereal at once – everything just gets clogged up and nothing comes out.

Goodbye, Praelexis

Fri, 14 Oct 2022 00:00:00 +0000

Today marked my last day at Praelexis, which has been my place of work for just shy of 6 years now. It’s been a wonderful journey, but it’s time for me to move on. Below I’ve attached my goodbye email that I sent out to the organization on my final day.

Hi there Praelexians,

It’s finally time for me to say goodbye.

Writing this email reminded me of one of my favourite panels in Berserk. In this scene two of the central characters, Guts and Casca, stare out over the Band of the Hawk’s campsite from atop a hill. Casca likens each little flickering campfire to each of the soldiers’ dreams and ambitions. She comments how all of the soldier’s have joined together to throw their own individual, little flames into the biggest bonfire of all, so that they won’t go out. Guts refers to this as the “campfire of dreams”.

Why your technical documentation sucks

Tue, 16 Aug 2022 00:00:00 +0000

This article was originally presented as a “brown bag talk”, an internal series of talks at the office where employees meet over lunch to share and discuss technical topics that interest them. It was also originally published as a post here, but I’ve slightly tweaked it to be a little more, ahem, opinionatedI’ll know I’ve peaked when I can say “fuck” on the company blog.. Enjoy!

In today’s post we’re going to (briefly) explore one of my favourite topics: technical documentation.If you’ve ever interacted with really good documentation, you’ll understand perfectly the kind of reverence that I have for it. More specifically, we’re going to look at some reasons why your technical documentation sucks. Along the way, we’ll also (sort of) touch on how we can make things better. The post is a bit hyperbolic and tongue-in-cheek, but the lessons aren’t — they remain useful whether you’re a team of one, or a team of one thousand.

Learning D3, Part One: Trial by fire

Fri, 31 Dec 2021 00:00:00 +0000

I’ve been putting off learning D3.js for a while now, primarily because I’ve isolated myself from fun distractions to complete my PhDA story for another day. But there will be one.. Alas, I’ve been on leave for a week, before I start full-time work in 2022. After 3 or 4 days of thesis-induced, delirium-riddled “rest”, I feel slightly renewed. At least enough to where I can face something quasi-adjacent to writing or programming, anyway. Not being one to sit idle for long, I’ve rolled up my sleeves and slowly started peeling back the layers of D3 as my motivation has ebbed and flowed.

On teaching

Mon, 20 Dec 2021 00:00:00 +0000

A month or so ago, I helped teach a three-day introductory Python course to client’s employees who were directly involved in, or adjacent to, credit modelling. Of the 8 or so people who attended, everyone was interested in learning Python. What made this a fun course to teach is that almost no one had previous experience with the language, only SAS. It was a challenge for me – teaching 8+ hours per day, three days in a row, is hard work. Anyway, I remember lying on the couch after the final day with a handful of takeaways from the experience, which I noted down and have shared here.

Adding videos in JupyterBook

Wed, 20 Oct 2021 00:00:00 +0000

A short post, hopefully to help out future fellow users who want to embed videos in their Jupyter Book.Book not Notebook. These are very different (albeit related) projects. It probably wasn’t the best idea to use this name, since googling for Jupyter Book only returns irrelevant results.

For a commercial endeavour, we’ve been using the Jupyter Book project to compile a bunch of instructional material (in the form Notebooks and Markdown) into an easy-to-publish format (static HTML) that is easy to distribute (the internet) and that looks beautiful (JupyterBook).

Covid Psychosis

Tue, 07 Sep 2021 00:00:00 +0000

Not many things have been able to break me from my from PhD-fueled writing hiatus, but I’ll admit people losing their minds isn’t a bad place to start.

This is based this on my observations of friends and coworkers as we all attempt to navigate the COVID situation (now that vaccines are available for over-18s in South Africa). It’s also heavily fueled by an excellent 94th episode of the Darkhorse Podcast, where Bret Weinstein articulates something I’ve believed for some time now. Right now, the world seems to be suffering from a kind of COVID psychosis, where the so-called “adults in the room” have seemingly lost touch with reality.

If—

Mon, 17 May 2021 00:00:00 +0000

If you can keep your head when all about you
 Are losing theirs and blaming it on you;
If you can trust yourself when all men doubt you,
 But make allowance for their doubting too:
If you can wait and not be tired by waiting,
 Or being lied about, don't deal in lies,
Or being hated don't give way to hating,
 And yet don't look too good, nor talk too wise;

If you can dream—and not make dreams your master;
 If you can think—and not make thoughts your aim,
If you can meet with Triumph and Disaster
 And treat those two impostors just the same:
If you can bear to hear the truth you've spoken
 Twisted by knaves to make a trap for fools,
Or watch the things you gave your life to, broken,
 And stoop and build 'em up with worn-out tools;

If you can make one heap of all your winnings
 And risk it on one turn of pitch-and-toss,
And lose, and start again at your beginnings
 ⁠And never breathe a word about your loss:
If you can force your heart and nerve and sinew
 To serve your turn long after they are gone,
And so hold on when there is nothing in you
 ⁠Except the Will which says to them: 'Hold on!'

If you can talk with crowds and keep your virtue,
 Or walk with Kings—nor lose the common touch,
If neither foes nor loving friends can hurt you,
 If all men count with you, but none too much:
If you can fill the unforgiving minute
 With sixty seconds' worth of distance run,
Yours is the Earth and everything that's in it,
 ⁠And—which is more—you'll be a Man, my son!

— Rudyard Kipling

Experimenting with Scrollytelling as Knowledge Share

Thu, 13 May 2021 00:00:00 +0000

“I’m in a hurry, just show me what you’ve made!” Take a look: The Buyer’s Persona, a whirlwhind tour.

I’ve been meaning to get back to one of my favourite pastimes: exploring different ways of telling stories using the web (especially when that story is involves data). Between finishing up my PhD (something I’ll write about soon enough, I’m sure) and building products at work, exploration has been relegated to the backseat recently.

The tricky thing about selling to businesses

Thu, 25 Feb 2021 00:00:00 +0000

For the past few months, I’ve been part of a team that’s been tasked with building out a product and selling it to businesses. Our primary target market is large corporations in the financial sector. This has been a unique challenge for the team – it’s near-impossible validating a product when it needs to be sold to a person who holds a managerial or department-head position. This kind of customer isn’t dime-a-dozen, so if you even get a meeting, the feedback is often heavily biased to that particular customer. I’m learning quickly how tricky this can be.

Who needs gradient descent? Getting to grips with Nevergrad.

Sun, 21 Feb 2021 00:00:00 +0000

Introduction

Nevergrad is an interesting and quite wonderful Python library that I’ve grown fond of over the last few months. It provides tooling to run numerical optimization without having to rely on computing the gradient of the function you’re optimizing (in stark contrast to the current algorithm du jour of the deep learning world, gradient descent). Think evolutionary algorithms, particle swarm optimization and so on.

I like a few things about nevergrad, and it’s since become my go-to tool for ad-hoc optimization problems, mostly due to how easy it has been to drop into my code. So, today, I’d like to talk about nevergrad for a bit.

Blisteringly-quick mocking when testing Jupyter Notebooks

Tue, 01 Sep 2020 09:09:00 +0200

Notebooks are good for two things::

Illustrative examples that provide code within the context of documentation (sometimes referred to as Literate Programming)
In-line plots.

For everything else, you should be using a “proper” Python module that has some tests – at least for the important bits.

This was the thesis of Joel Grus’ hilarious talk “I don’t like notebooks”, which I happen to largely agree with. And I think it’s excellent advice…

Building a product website in half a day with no web dev knowledge

Tue, 18 Aug 2020 09:00:00 +0200

I had an interesting challenge at work a month or two ago. It went something like this, “Michael, while we’re waiting for the official website to get rebuilt, it would be great if you could build a website for our product as an alternative. Spend a day on it, let’s see what you can get.”

I love this kind of task or challenge.

I’m going to talk you through the process of how I did exactly that. I unfortunately can’t show the final result (since the product hasn’t officially launched yet – nor is it guaranteed we’ll be using this rapid-fire website), but I’ll be giving some general advice and my approach for getting a great-looking product page up fast.

Lean Startup summary notes + some closing thoughts

Wed, 03 Jun 2020 09:00:00 +0200

Over the last few days I finished reading and compiling my notes on The Lean Startup by Eric Ries. The summary notes are available from the Learning page.

I really enjoyed the book. I particularly like the build-measure-learn feedback loop, and the concept of small batches. This has the potential to be tremendously influential in how I conduct my projects and work. I feel naturally drawn to hypothesis-driven development, and Lean Startup seems to provide a decent framework to execute this idea.

A short word about comments in code

Tue, 19 May 2020 22:40:28 +0200

A common fallacy is to assume authors of incomprehensible code will be able to express themselves clearly in comments.

— Kevlin Henney

I quite like Kevlin Henney, particularly his talks. Some people grumble about his information density being a little bit low, (as if optimizing for information density is a good proxy for optimizing for information transfer – it’s not), but I find his talks both informative and entertaining. Above all though, I find them accessible. Kevlin spends a lot of time talking about is the mindset of clean code, which we’ll briefly encounter today.