This site is an extended resume — a working portfolio of directive frameworks, agent loops routed by logic gates and minimal-surface decision boundaries, and full-stack platforms I have built end-to-end. The other sites linked from here (KensGames, TTLRecall, the dimensional-programming writeups) are personal portfolio projects demonstrating the same skill set in different domains. None are commercial products, businesses, or services for sale.
A point becomes the whole. The whole becomes a point in a higher dimension.
This is the architectural principle I developed across the portfolio projects below. The butterfly effect taken literally — not "small thing causes big thing" but small thing IS big thing IS small thing, depending on which octave you address it from. A wing-flap is a point. A hurricane is the whole. A century of climate data is the hurricane as a single point. Same substrate, three altitudes.
In code: a seed x is folded through a lens y to produce a
delta z; the closed manifold m becomes the seed
xn+1 of the next octave. D7 of one breath is D0 of the next.
The substrate is self-similar at every scale. I have used this discipline to keep
AI directives stable, route agent traffic deterministically, and structure data
stores around geometry instead of bespoke tables.
The other personal portfolio sites linked from here each apply this framework to a different domain — KensGames (the substrate at play), DimensionalProgramming.com (the substrate formalized as a writeup), and several others in progress. They are demonstrations of skill, not products for sale.
I write the reasoning frameworks that keep AI systems stable. Directives are structured prompts plus governing rules — an algebra of allowed transformations that prevents drift, hallucination, and brittle failure modes. Demonstrated across the portfolio projects on this site.
Read a full directive →I design agents whose job is to make other AI work better — verifier loops, evaluation harnesses, retrieval-augmented pipelines, and rule engines that catch what a single LLM call cannot. Practical orchestration over hype.
How I design agents →Truth tables, logic gates, and decision trees give an AI system a hard skeleton it cannot lie around. Manifolds — geometric surfaces governed by z = x · y — give it a way to focus attention on the small region of state that actually matters. Together they replace guesswork with geometry.
How I think →Route by regex pre-filter, truth-table constraints, decision-tree shortcuts, and a Schwarz-D minimal-surface boundary so frontier credits go only where frontier reasoning is required. The router is real, tested, and live on this site.
Schema-validated outputs, retry budgets, verifier ensembles. LLMs never return free text to the next stage. Drift dies at the boundary.
No destructive updates. Every state is traceable through its lineage on the underlying manifold — debugging becomes walking the path, not guessing the past.
I own the VPS, the deploy pipeline, the data model, the directive doctrine. Vendor portability is a first-class requirement — no primitive I cannot replace within one sprint.
Each link below is a personal portfolio piece — built end-to-end to demonstrate the skills above. None are commercial products or services for sale.
3D game platform with deterministic simulation, real-time rendering, authentication, and multiplayer orchestration. Designed and built solo as a portfolio piece.
Rapid-build challenge platform with AI-accelerated workflows for scenario generation, scoring, and feedback. Personal portfolio project.
A technical writeup site presenting the paradigm I developed — z = x · y as a substrate, breath verbs, the manifold-as-database.
Public projects and code samples across AI tooling, agent design, and dimensional programming.