Every project documented here is published and available to review. No demos, no slide decks, no pitch — the work itself is accessible at the links below.
A complete AI-assisted adaptive learning platform built on a formally documented pedagogical framework. The system diagnoses the exact schema failure responsible for a student's errors — not the surface error but the upstream conceptual corruption that produced it — and generates problems calibrated to the student's specific cognitive state. Every session updates the student's Dynamic Learning Profile, making the system more precise over time.
This is not a product with a pedagogy layer on top. The pedagogy came first — seven years of direct classroom observation — and the system was designed to implement it. The architecture includes tier maps (concept dependency graphs), a misconception ID system, a 80/20 calibration engine, and an adaptive difficulty mechanism. The full technical specification is published.
Eleven formal papers documenting each principle of the Lacefield Pedagogical Framework — with the mechanism behind it, the evidence base, the specific misconceptions it addresses, and the classroom observations that originally produced it. Each paper is self-contained and links to the others.
The papers cover: performance vs. understanding, reading as mathematical foundation, productive struggle calibration, foundational fluency and cognitive load, precision of definition, incorrect correction, confidence as an educational variable, perseverance over aptitude, active recall over passive review, mathematics as metaphysics, and the intake diagnostic and Dynamic Learning Profile.
Ten formal concept dependency graphs covering the full GED curriculum plus standalone algebra, geometry, fractions/decimals/percents, grammar, and economics. Each node documents the concept definition, strict prerequisites, specific misconceptions with IDs, fluency requirements, lexical ambiguity risks, diagnostic probe questions, and real-world context hooks.
This is the most granular public documentation of mathematical concept dependencies for GED through calculus currently available anywhere. The tier maps are the data layer that makes adaptive problem generation possible — the system reads the student's schema floor and generates problems targeting the specific misconceptions flagged in their profile.
Original correlation analysis conducted on 130+ TABE placement scores during GED instruction in Florida's correctional system. Finding: reading comprehension predicts applied math performance more reliably than language arts scores — which is counterintuitive given conventional grouping of reading and language arts together.
The mechanism is straightforward: the GED math section is primarily word problems. Before a student can apply any mathematical procedure, they must read the problem precisely enough to identify what is being asked. Most math errors in word problems are reading errors that happen before the first calculation. The finding has direct implications for how math diagnostics should be structured — reading baseline should come first, not as a separate assessment.
Formal write-up in preparation for peer review submission. The existing white paper documents the finding and methodology.
A long-form referenced essay making the case for what adaptive learning could be if built from cognitive science rather than engagement optimization. Covers Bloom's 2-sigma problem, cognitive load theory, spaced retrieval research, zone of proximal development, and the evidence base underlying every principle in the framework. Written for an intelligent general audience, not a specialist one.