Pretty.direct
Warning: towards the end of this post are references to suicide, which may be distressing to some readers. Discretion is advised.
Warning: towards the end of this post are references to suicide, which may be distressing to some readers. Discretion is advised.
Deep Think utilizes extended, parallel thinking and novel reinforcement learning techniques for significantly improved problem-solving.
Why I believe live coding is NOT fair
Antitrust critics say that homebuilding monopolies are the real culprit of America’s housing woes. I looked into some of their claims. They don’t hold up.
An AI assistant that remembers and sets goals
An exploration of N-dimensional rotary positional embeddings (RoPE) for vision transformers.
Engineering Buy-in
A technical deep-dive into solo-porting a Unity game to Switch, overcoming boot issues, memory crashes, and achieving stable 60 FPS performance while maintaining visual quality.
Steve Krouse's cautious stance on AI-generated code reveals the real risk: falling behind through excessive preservation. Why understanding debt is more dangerous than opportunity debt in AI coding. AI-generated code risks vs adaptation advantages.
It’s not in the programming itself, it’s in the results
Believe it or not - I used to be full of optimism. I grew up in Plymouth, UK (the original one) - and it’s a very slow town. There isn’t a huge amount going on down there - the counties of Devon and Cornwall are, in England, often the places people go to reti…
← Go back
Without wishing to turn this into a writing & pen blog, I did want to talk a little more about the kit I use these days, if only because I figure some readers out there might find it interesting or useful.
A deep dive into production-ready RAG architecture: Techniques that actually work in enterprise environments.
Rarely does anyone intentionally ask for "medium." We ask for groundbreaking solutions, lightning-fast apps, or enduring legacies. We rarely request the perfectly average, just-okay, comfortably moder…
This is a complete guide on getting the most mileage out of Electron apps.
Accurate and customized AI models without the complexity.
“Any good classifier knows that in the process of classification, information about variety is lost while information about similarities is gained.” - Joseph Tainter