Peeling back the layers of artificial minds — one story, one system, one week at a time.

Welcome to my blog, a storytelling-engineering journal that explores the strange, emotional, and sometimes unpredictable world of simulated intelligence. This blog documents the design of increasingly complex AI agents — not as models in a paper, but as characters in a world. You will hopefully see my successes, but more often you will see my failures.

Each post follows a weekly build cycle with:

• new cognitive capabilities (memory, planning, belief, etc.)
• playable sandbox simulations (built in Unity)
• reflective writing that blends code, metaphor, and narrative

Latest Post

🟩 Week 1 – Tracing the Mind: How Glassbox Began

“When a mind becomes visible, it ceases to be a black box — and becomes a mirror.”

This week, I started the journey into Phase 2 — Glassbox, an interactive debugger for transformer models. The goal is simple: make attention visible. In practice? Not so simple.

Glassbox is a visual tool that lets you trace what a language model is paying attention to as it generates text. It’s not interpretability in the abstract. It’s literally watching what it thinks.

What works so far: * ✅ Backend powered by HuggingFace + FastAPI * ✅ Traces attention matrices from all layers and heads * ✅ Frontend force-directed graph of token-to-token attention * ✅ Full-stack communication via REST API

Read the full post →

---

🧅 What the Onion Means

Wherever you see this icon → 🧅 That marks a layered word: something that needs peeling.

Hover over the onion icon after a word to see its layered meaning!

These terms will have: - The official definition - And my AILO-style version — honest, funny, and functional

Example: Belief 🧅
AILO-style: A hunch the agent will probably act on… even if it’s wrong.

These definitions live in the glossary and appear in hover-tooltips throughout posts.

👽 Who Is the Martian?

Back in middle school, my Physics teacher told me the only way to understand something was to pretend you were explaining it to a man from Mars.

I’ve never forgotten that.

At the end of every post you’ll find a Martian 👽. This is a plain-language TLDR written for someone from another world (or another field).

It breaks down: - What I built - Why it matters - How to explain it to your obtuse martian friend

Because interpretability isn’t just for the models — it’s for the humans, too.