Rebuilding this site (and how writing works here now)

This site is a pile of Markdown files now. The old version was a React app, and adding anything meant editing components. By the time I finished fiddling with JSX, I didn’t feel like writing anymore. The new one is Astro rendering static pages. Every post, including this one, is a note dropped into a folder and pushed.

I write in Obsidian, so the pipeline is built around it. A file needs zero frontmatter: the title falls back to the first # heading (or the filename), the date falls back to the file’s first git commit, and the description gets lifted from the opening paragraph. Wikilinks, embeds, highlights, and callouts all work.

The 10-second publishing loop

Write a note in Obsidian → drag the .md into site/src/content/blog/ → git push. GitHub Actions builds and deploys the rest.

Things a research blog should render

Math, because the PPO clipped objective shows up in basically everything I do:

L^{CLIP}(\theta) = \hat{\mathbb{E}}_t\left[\min\left(r_t(\theta)\hat{A}_t,\ \text{clip}(r_t(\theta), 1-\epsilon, 1+\epsilon)\hat{A}_t\right)\right]

Code, with proper highlighting:

def masked_policy_step(obs, mask, policy):
    """Physics-aware action masking, the one-liner version."""
    logits = policy(obs)
    logits[~mask] = -float("inf")   # invalid actions never get sampled
    return torch.distributions.Categorical(logits=logits).sample()

And highlights, footnotes¹, and callouts for the load-bearing caveats:

Reproducibility disclaimer

If a plot in a post disagrees with the paper, trust the paper. Posts are lab notes, not camera-ready.

Why bother

Papers are where results end up, but most of research happens between papers: negative results, tooling, half-formed intuitions. I wanted a place for that stuff. If any of it is useful to you, or wrong, email me.

Footnotes render like this, tucked at the bottom. ↩

Things a research blog should render

Why bother

Footnotes