
Trip to Southern Utah
We flew to Las Vegas, drove into southern Utah, and hiked around Zion, Escalante, and Bryce Canyon.
Notes and essays
Software, AI tools, technical books, and life outside work.

We flew to Las Vegas, drove into southern Utah, and hiked around Zion, Escalante, and Bryce Canyon.

When a coding agent makes a repeatable mistake, I try to keep the correction in the harness for the next run.

My first week in San Francisco: paperwork, an unfurnished apartment, two cats, Bay Trail runs, and waiting for furniture.

A browser-based load-flow solver and Mandelbrot explorer let me revisit power-system analysis, numerical precision, and the limits of small client-side tools.

A concrete way to understand dropout in deep learning: sampled subnetworks with shared weights, standing in for a much more expensive ensemble.

Notes on the 11-inch M3 iPad Air for reading, handwriting, and remote access without turning it into a laptop replacement.

I replaced scattered responsive-image conventions in a Next.js static site with one generated manifest for variants, dimensions, and lookup during static rendering.

A lightweight intake form, spreadsheet cleanup script, or dashboard mock can make an internal workflow concrete before it needs production engineering.

Codex helps me make small site edits from my phone; Sora remains a way to test visual directions before I know whether they belong.

Chapter 6 separates representational capacity, parameter efficiency, and learnability—and shows why depth changes scaling, not just expressiveness.

Part I of Goodfellow, Bengio, and Courville works best as shared vocabulary for later chapters, with proof details still worth supplementing.

My AI-assisted coding workflow now uses short briefs, plan reviews, checkpoints, and build/lint/test verification.

How repo-backed Markdown becomes static HTML without a CMS, database, or runtime admin surface.
Occasional updates when I publish something new.
Follow.it sends the emails and includes an unsubscribe link.