Design, cost, secure and prove AI systems a client can sign off on.

Fifteen lessons and four labs that take one engagement from a vague wish to a production system: discovery and non-functional budgets, model and retrieval and agent decisions recorded as ADRs, the evidence pack (evals, cost model, threat model), and the runbook, compliance annex and rollout plan that let it go live and stay live.

Start the free lesson →See the full curriculumPricingLabs
Lessons
15
Project labs
4
Modules
4
Reading
~9h

You finish holding something.

You leave holding an Architecture Dossier for a real engagement, built one file per lesson: charter, budgets, decision records, diagrams, eval plan, cost model, threat model, compliance annex, runbook and rollout plan. It is the document a buyer signs and an engineering team builds from.

The artifact

The Architecture Dossier: a folder of decision records, budgets, diagrams, an eval plan, a cost model, a threat model, a compliance annex, a runbook and a staged rollout plan for one AI system, plus a one-page executive summary.

The loop you learn

Frame → Decide → Prove → Run

Every lesson has the same shape.

You know what you are buying before you buy it.

01The failure

A specific system that broke, with the symptom and what it cost, before any term is named. You commit to a decision before reading on.

02The rule

The mechanism explained with a diagram, then the decision stated as a rule with its condition and a contrast table of the alternatives.

03The cases

A worked case with the reasoning shown step by step, a faded case where one decision is yours, and an open case the quiz grades.

04The deliverable

One named file added to your Architecture Dossier, stated identically in the outline, the lesson header and the closing section.

Why it is built this way.

Three decisions that shaped every lesson.

Decisions, not catalogues

No lesson lists a vendor's services. Every lesson teaches one decision as a rule with its condition, then makes you take it on a case with real constraints.

Evidence before production

A model choice, a retrieval design or an agent pattern is an opinion until an eval, a cost model and a threat model say otherwise. Every design decision here ships with the evidence that proves it.

Numbers carry a source

Prices, limits, dates and benchmark figures rot in months. Every precise number in this course links to where it came from and when it was checked, or is marked as an illustrative composite.

4 modules, 19 entries.

The whole path is public. Nothing is hidden until you pay — only the bodies are gated.

MOD 01Frame: the engagement

Turn a wish into a chartered engagement with budgets: what the role decides, how discovery produces numbers, and how the client's segment fixes the constraints before any design starts.

MOD 02Decide: the system

The six design decisions every LLM system contains, each taken by the budgets and recorded as a decision record: model, context, grounding data, agent pattern, integration boundaries, and where it runs.

MOD 03Prove: the evidence

The three documents that turn a design into a defensible pilot: an evaluation plan with a validated judge, a cost model the CFO can audit, and a threat model that treats the agent as the attack surface.

MOD 04Run: production and governance

What lets the system go live and stay live: obligations mapped to design elements, a runbook with SLOs and fallbacks, and a staged rollout whose gates, human oversight and exit options are written down.

Start with the free lesson.

Senior engineers, tech leads and consultants who must design, cost, secure, evaluate and operate LLM and agent systems for clients, and defend those decisions to the people paying for them.

  • One-time price, lifetime updates
  • Every course included — this one and the next
  • 30-day money-back guarantee
  • No card required to start
Open lesson 1 — freeSee what unlocking costs