Skip to content

Blog

Technical writing on AI, agentic engineering, and systems I built. Code included.

Software Will Commoditize

The cost of software converges with the cost of the energy needed to produce it, and most of the enterprise software industry hasn't noticed yet.
August 10, 2026

The Free Banana Problem

Why losing all our jobs to AI could be a good thing, and what has to go right during the transition.
July 14, 2026

Why the Suits Aren't Seeing ROI on AI

The people signing off on AI architectures have never run Claude Code. The middle layer is where the great filter happens.
July 10, 2026

Just Apply Determinism

Agents work best when you give them fast, reliable feedback. Here are the tools I use to do that.
July 6, 2026

Training in Orbit

What happens when you try to train a neural network on a GPU that's getting hit by cosmic rays. Turns out the fix is surprisingly simple.
March 27, 2026

The Bias Ledger

An LLM lost 4.7% during COVID while the market lost 34%. Then it got fooled by a dead cat bounce in 2022. Both results support the same thesis.
February 7, 2026

Closed Loops

How agentic programming changes the fundamental unit of engineering work, and why TypeScript matters more than I expected.
January 30, 2026

Scaling

Notes on The Scaling Era: training scale, post-training, and test-time compute as three interest rates compounding together.
October 31, 2025

Traffic Lights Considered Harmful

The average driver spends two to three months of their life at red lights. I simulated what happens when autonomous vehicles negotiate intersections directly: 60% less waiting, and a verification gap I don't want to minimize.
August 5, 2025

The Intelligence Age

Sam Altman wrote The Intelligence Age. I want to believe the optimistic version, but it skips too quickly past the hard parts. Hopeful about the destination, worried about the path.
September 27, 2024

Attention

How attention works in transformers, with the math. This is the mechanism behind models like ChatGPT.
June 28, 2024

Agents

An agent is an LLM running in a loop. What makes the loop work, where it breaks, and how far it scales.
June 5, 2024

The End of Gas

When does buying a gas car become obviously wrong instead of arguably suboptimal? The math says sooner than most people think.
January 5, 2024