This is the first post on my blog. I’ll use this space for working notes and essays on the things I spend my days thinking about: agentic reasoning, inference-time compute, and large-scale neuroevolution as a path toward AGI.

This post doubles as a template — it shows every kind of content the blog supports, so future posts are copy-paste-and-edit.

Text, links, and emphasis

Standard Markdown works as you’d expect: bold, italic, inline code, and links. Paragraphs are set at a comfortable reading measure with generous line height, matching the rest of the site.

The thesis in one line: pre-training scaling is hitting diminishing returns, and the next leg comes from recursive self-improvement at inference time — gradient descent inside, evolutionary search outside.

Images with captions

Drop an image into img/blog/<post-slug>/ and reference it with the figure include for a captioned, responsive image:

An outer evolutionary loop wrapping an inner gradient-descent loop.
Neuroevolution as an outer-loop substrate: gradient descent trains each network, evolution searches over architectures.

Video and media

Embed a talk or demo — YouTube or a self-hosted clip — with the video include:

Code

Fenced code blocks get syntax highlighting automatically:

def evolve(population, generations):
    for _ in range(generations):
        for net in population:
            net.train()            # inner loop: gradient descent
        population = select(population)  # outer loop: evolution
        population += mutate(population)
    return best(population)

Lists

  • Write posts in plain Markdown
  • Commit and push — GitHub Pages rebuilds the site automatically
  • The post appears at /blog/<title>/ within a minute

That’s the whole workflow. More soon.