Blog
AI Isn't Why You Lost Your Coding Job
It was the abstraction all along. We never check the compiler's output, and soon we won't bother to check the generated code either.
Build It for One Person
Shipping a new app every week and hoping one works out is a gambler mentality. Build something one person will absolutely love instead.
Better Math = Better AI?
OpenAI just released 722 math papers written by an internal model. What was claimed, how much is verified, and whether AI-discovered math could be what builds the next, better AI.
Run Your Own Race
Some people will always start ahead of you. You can accept that as information and still keep running from wherever you started.
Write Like an Aircraft Manual: ASD-STE100 for LLMs
A 1980s aircraft-maintenance standard is now a one-line fix for wordy model output. What it is, why it works, and what it costs.
Two Kinds of Engineers in 2026
AI will do everything, or AI can do a lot and I can now do things I could not before. The difference between those two sentences decides whether you keep finding problems worth solving.
Why AI Needs to Know What It Doesn't Know
Models got much better at answering in 2025. The next step is metacognition: knowing when they might be wrong, and saying so.
Contrastive Language Models, Explained: How CLM-8B Chooses Actions Without Writing Text
What a Contrastive Language Model is, how CLM-8B is built and trained, where its speed comes from, and where it falls short.
Finally, Graduated
Four years, a lot of chaos, a lot of learning, and one chapter that is officially done.