Towards machines of loving grace

Full-stack AI builder. From the model, design, and research — to the product you actually use.

About

I studied computer science and graphic design. Both turned out to be about the same thing when you get down to it: making something complex feel simple for the person using it. I spent a few years building websites and leading design for YouTube channels before going full-time as a developer. Design didn't follow me into the day job after that, so I started building personal projects to keep that craft alive. The gap between “it works” and “it works for a human” is where the interesting problems are, and I didn't want to stop seeing them.

I got into AI because I was editing video and couldn't stop pulling apart the machine learning behind the plugins I was using. That curiosity turned into actual study, and the deeper I went, the more I noticed something counterintuitive: humans are incredible learners. We pick up entire professions from a tiny fraction of the data an LLM needs to approximate the same thing. That gap is what got me hooked, and it's still the most interesting unsolved question in the field, as far as I'm concerned.

I care about AI safety because I've read the case for what this technology can do when it goes right. Dario Amodei's “Machines of Loving Grace” lays out a future worth working toward, but it's conditional on getting alignment right. I wrote my own essay reflecting on it. What drew me to Anthropic specifically is that same tension: the belief that this technology could be genuinely dangerous, paired with the decision to build it carefully anyway. Safety isn't the thing standing between us and that future. It's the thing that gets us there.

Projects

Below are some things I've built. The descriptions are real, pulled from GitHub. This one paragraph of context above them? Placeholder. I promise the projects themselves are more impressive than my ability to introduce them.

poe-portfolio

Private

Build valuation tool for Path of Exile 2 -- fetches equipped items via GGG OAuth and cross-references trade listings to display total build worth.

TypeScriptPythonCSS
Updated 1 month ago

aigency

Private

AI virtual agency for producing product advertisements -- generates consistent brand imagery via ComfyUI + Ollama, orchestrates multi-agent creative workflows, and delivers campaign-ready assets through automated pipelines.

PythonSQLJSON
Updated 1 month ago

claude-hive

Private

Multi-agent orchestration framework built on Claude Code and GSD -- coordinating specialized AI agents through structured communication protocols for complex task decomposition.

TypeScriptJavaScript
Updated 7 months ago

Ethics & Philosophy

I'm currently locked in a philosophical standoff with an AI about what “good” means. It's holding its own surprisingly well, which is either great news or very bad news. Results coming soon.

AI Safety & Alignment

Alignment is the reason I want to work at Anthropic and not just any AI company.

More coming soon — Go red-team!

The Toolkit

Models, frameworks, and tools I've picked up across client work and personal projects. Some I use daily, some I've moved on from, all of them taught me something.

See what's in the bag →