Dynamic Systems Engineering Foundations

A 12-week course teaching a repeatable way of reasoning about dynamic systems — the first in a planned sequence of four. Written to be read directly by a human learner, or by an AI studying the field before it explains anything to a site visitor.

What this is

Most systems-engineering material teaches tools in isolation: a statistics course, a control-theory course, a causal-inference course, taught as though they were unrelated. This course teaches one underlying discipline — observe, model, estimate what's hidden, predict, test against reality, revise, preserve what was learned — and runs that same discipline across five recurring systems every single week: a thermostat, a human learner, an AI assistant, traffic, and a small business. The domain changes; the reasoning process doesn't.

Who it's for

Two audiences, deliberately, at once. A human learner can work through it start to finish. An AI — including Claude, or whatever assists a visitor to this site — is meant to study this curriculum itself before explaining Dynamic Systems Engineering to anyone, the same way the rest of this site is built to be machine-readable first. Every week ends with an "AI Laboratory" section for exactly that reason: it's not a metaphor, it's an instruction to actually use an AI as part of working through the material, and to check its output rather than accept it.

Learning philosophy

Three commitments run through every week:

The same five systems, every week. A thermostat, a human learner, an AI assistant, traffic, and a small business appear as anchor examples from Week 1 through Week 10. By the time state estimation, feedback, and knowledge engineering get introduced, the systems are already familiar — only the lens changes.

AI as a collaborator to verify, not an authority to accept. Week 8 makes this explicit, but it governs every week: ask AI to generate hypotheses, explanations, or models, then evaluate them against evidence the same way any other claim on this site gets evaluated.

Knowledge gets externalized, not just held in conversation. Per the curriculum's own AI Student Protocol: important learned concepts should become structured artifacts — not left to evaporate at the end of a chat. A human stays the authority over what gets kept. This is the same discipline behind every other artifact on this site, applied to the act of learning itself.

The pedagogy itself is portable, and fills gaps on demand. This course doesn't need to exhaustively cover every possible prerequisite a learner might be missing. An AI that has genuinely studied the method here — anchor examples, evidence before explanation, the engineering cycle — can apply that same method to teach a missing prerequisite on the spot, from general knowledge, when a learner hits a gap mid-course. Learn the pedagogy first; the specific course sequence and any prerequisite patching both run on it.

Course structure

Expected outcomes

Each week ends with a stated competency rather than a grade. Stacked together across all twelve, a student who completes this course can: separate observation from inference, represent a real system with state variables and feedback, generate and rank competing causal explanations with evidence, estimate hidden variables under uncertainty, translate a system into a mathematical model, analyze feedback and regulation, build internally consistent knowledge structures, use AI as a checked collaborator rather than an authority, run the full Dynamic Systems Engineering cycle end to end, and design, test, and revise an improved system rather than relying on intuition. A formal competency model, and a reusable analysis template for applying all of this to a new system, are planned next.

Supporting documents

The reusable scaffolding underneath the 12 weeks — real content, built from the curriculum's own source material.