Blog Posts
Notes on AI, infrastructure, distributed systems, and whatever I'm building this week.
A day with project embeds: rotating creds, scrubbing history, and replacing static bearers with OIDC
The previous post sketched an architecture for hosting live, interactive versions of every project on a personal site. By the time I'd onboarded the third project, three things broke that the first sketch hadn't…
Build anything, make it playable: an architecture for discoverable project embeds
A personal site that lists projects is fine. A personal site where you can actually use the projects — type input into a compiler, query a database in the browser, play against an old game-playing agent, watch a graph…
A reference gallery: every Mermaid diagram type the blog supports
A live reference of every Mermaid diagram type this blog supports. Each block below is rendered through the same build-time pipeline as every other diagram on the site: rehype-mermaid + Playwright at SSR, mermaid's…
Theming Mermaid diagrams properly: a five-PR debug arc, and what actually fixed it
Five days ago I shipped Mermaid rendering on this blog. That post claimed:
Diagrams use Mermaid's neutral theme so they read cleanly on both light and dark backgrounds. (We can't follow the manual theme toggle without…
Visualizations now render: Mermaid diagrams + images on the new blog
import { Image } from 'astro:assets'
Update — 2026-05-27: the "Theming" section below is now historical. The neutral-theme-everywhere claim turned out to be wrong: mermaid's SVG-baked <style> element painted white…
SOTA ML by July 2025: From Explicit Features to Implicit World Models
Thesis: Modern ML keeps rediscovering the same north star: better compression of the data-generating process yields better generalization and capability. The path there has shifted from explicit, hand‑crafted features…
Accelerating the Future with Llama 4 on SageMaker JumpStart
Below is a comprehensive blog post that covers the release of Llama 4 on SageMaker JumpStart, structured with an overview of implementation details, key features, and best practices. This blog builds on information…
Capturing Discord Calls with Rust: Building the Foundation for AI Note-Takers
In today’s world, our fast-paced lives rarely allow us the luxury of remembering every detail from important meetings—or those long Dungeons & Dragons sessions. As a multi-tasker with a notoriously poor memory when it…
Beyond Commands: A Vision for Voice Assistants That Evolve into Plugin-Installing AI Agents
The digital assistants we interact with today—like Alexa and Google Assistant—offer a glimpse into a convenient, automated future. However, their capabilities remain limited by rigid command structures and a shallow…
How to Build an AI Agent That Writes Terraform for You (and Knows Your AWS Setup)
A comprehensive conceptual blueprint for a blog post that combines practical guidance on developing an agent that intelligently generates Terraform code with insights into leveraging your AWS account’s structure and…