Making meaning from messy data
Every project here starts with something genuinely messy — fourteen years of workouts, four years of hive photos, a family's stories, a real LLC's books, 426 anime subtitle files — and ends with structure you can query, verify, and trust. Built in public, documented end to end, running live wherever possible.
Prefer to explore visually? Open the interactive project map →
The Meaning Layer
A 12-week semantic web curriculum — RDF, OWL, SPARQL, through to a deployed hybrid LLM + knowledge graph capstone — published as I work through it, wrong turns included. Two of the projects above are its anchor canvases: the Naruto network carries the ontology modeling, and my resume graph makes the case for linked data concrete.
PROGRESS.md · updated weekly · module 1 of 4 in progress
The workbench
Smaller builds and earlier work — kept here because the through-lines are visible in hindsight.
What the cognitive science actually does here.
My dissertation was about visual attention — how people decide where to look. I won't claim cognitive science blueprints every system on this page; it doesn't, and dressing engineering decisions up in borrowed neuroscience is exactly the kind of overselling I distrust.
What the training actually left me with is a set of stubborn questions I bring to every messy dataset: What deserves attention? What would count as evidence? Where does the meaning live? A few projects wear those questions openly. In most, they just shaped a hundred small decisions — and the receipts above are how you can check my work.
If you're building AI systems people need to trust — or sitting on messy data that should mean something — let's talk.
Work with me Read the writing