The Engineering Series · 6 Courses · 1 Platform
// From first API call to docker compose up.

The AI Systems Architect curriculum.
One project. One running platform.

161
lessons
6
courses
1
running platform
Get on the waitlist
€797 waitlist
Self-paced · 14-day conditional refund · Apache 2.0 code
The gap

There is no AI Systems Architect curriculum.
So I built one.

The market is full of AI courses. They all teach the same thing. Call an API. Chain some prompts. Build a chatbot. Deploy to Vercel. Screenshot it for LinkedIn.

None of them teach you how to build AI systems that a CISO will approve, a general counsel will sign off, and a VP of Engineering won't lose sleep over.

I looked for this curriculum when I started building production AI systems for enterprises. It didn't exist. After 60+ systems for PwC and the most demanding companies, I built it myself.

Six courses. One commit history. The portfolio piece that proves you can build the real thing.
// engineering-series.manifest

The signed inventory of what the series ships.

One project carried across six courses. Every artifact is yours: the code, the deployment, the documentation.

courses6 (E1 → E6)
research3 arXiv papers + 1 book
lessons161
code6 OSS repos · Apache 2.0
modules35
capstone4-layer Enterprise AI OS
pace41–52 weeks self-paced
deploydocker compose up
// status: deployable
The six courses

Each course adds a layer.
By E6, you have a running platform.

The same project carries across all six. Versioned API in E1. Trust layer in E2. Governance middleware in E3. Security hardening in E4. Context pipeline in E5. Full platform in E6.

Course E1 01
"Stop prompting. Start engineering."

AI Engineering Foundations

LLM API architecture. Prompt architecture (not "prompt engineering"). Tool use via MCP. The GRAIL Loop in code. Evaluation-first development. Your first production AI service.

See course →
29 lessons€197
Course E2 02
"Your system must know when to say 'I don't know.'"

Trust Engineering

Self-consistency sampling. Conformal prediction (from CM's arXiv paper). TrustGate integration. Prediction set routing. Behavioral drift detection. The math that makes AI trustworthy.

See course →
30 lessons€197
Course E3 03
"Governance is not a feature. Governance is the platform."

Building Accountable AI Agents

The 4-layer AI OS architecture. 7 Platform Services: auth, guardrails, actions, context, gateway, monitor, audit. The Platform Protocol (~15 API calls). The system that escapes Pilot Purgatory.

See course →
26 lessons€197
Course E4 04
"Break your own systems before someone else does."

AI Security Engineering

The 12 prompt dimensions from CM's research. Prompt injection (direct, indirect, goal reframing). Red-teaming framework. Defense architecture. Guardrails implementation. The hardened system.

See course →
21 lessons€197
Course E5 05
"The document is the AI OS layer, engineered."

Context Architecture

The Context Pipeline: audit, classify, store, route, filter, serve. Permission-filtered context. Beyond RAG. Context Kubernetes for enterprise-scale knowledge orchestration. The discipline the industry just named.

See course →
29 lessons€197
Course E6 · Capstone 06
"Assemble the full platform. Ship it."

The Enterprise AI OS

Every component from E1-E5 assembled into one running platform. Add the missing layers (organisational agents, intelligence systems, desktop shell). Deploy with docker compose up. The capstone of the series.

See course →
26 lessons€197
The capstone build sequence

The capstone you finish in E6 is the project you started in E1.

Every course adds a layer. Nothing is thrown away.

E1
Foundation
Versioned prompts. Multi-model. MCP. GRAIL eval. Logging.
+ E2
+ Trust
Self-consistency. TrustGate. Reliability guarantees.
+ E3
+ Governance
7 services. Platform Protocol. ~15 APIs.
+ E4
+ Security
Guardrails. Agent Auth. Sandboxing. Red-team tested.
+ E5
+ Context
Multi-source. RAG. Context Router. RBAC.
= E6
Full Platform
All 4 layers. Org Agents. Intelligence. Desktop Shell.
// One commit history. Every artifact still in the repo at the end.
Backbone

Built on published research and open-source code. Not blog posts.

Built on CM's arXiv papers, formal proofs and open-source implementations.

// research

arXiv · 2602.21368
Self-Consistency & Conformal Calibration
Distribution-free reliability guarantees. The math behind TrustGate. Foundation of E2.
arXiv · 2604.04561
Mapping the Exploitation Surface
What makes coding agents seek a planted backdoor. Foundation of E4.
arXiv · 2604.11623
Context Kubernetes
Declarative context at enterprise scale. TLA+ and Alloy formal specs. Foundation of E5.
Book · 130+ pages
The Enterprise Agentic Platform
The 4-layer AI OS. 7 design principles. Platform Protocol. 90-day plan. Foundation of E3 and E6.

// open-source · Apache 2.0

trustgate
Reliability certification. Conformal prediction.
guardrails
Policy enforcement. I/O scanning. YAML rules.
context-router
Intent-based retrieval. Permission filtering.
agent-monitor
Observability. Behavior monitoring. Anomaly detection.
agent-auth
Identity. Permissions. 3-tier approval. SSO.
context-kubernetes
Declarative context. Formal specs. Reconciliation.
Pricing

All six courses. One purchase. Lifetime access.

Save €385 vs. individual
€1,182 individually
€797
One-time payment. Lifetime access. All future updates.
  • All 6 engineering courses (E1 through E6)
  • 161 lessons across 35 modules (video, written, runnable code)
  • 6 capstones building into one running 4-layer Enterprise AI OS
  • 3 arXiv paper walkthroughs and 1 book (130+ pages)
  • All 6 Cohorte open-source repos (Apache 2.0)
  • docker compose production deployment, runbook, monitoring config
6 months in the Engine Room. Where alumni and operators go to get unstuck. Twice the access of a single course.
Get on the waitlist
14-day refund if you haven't completed the first module. After that, the series is yours.
FAQ

Before you ask.

Anything else? Email [email protected].

Why €797 vs €1,182 (6 × €197)?
The bundle is the recommended purchase path. The €385 saving reflects that the courses build on each other — buying them piecemeal forces you to backfill context. The bundle is also the only path with a 14-day conditional refund.
Can I take them out of order?
Not recommended. Each course adds a layer to the same project: the Enterprise AI OS you finish in E6 is the one you started in E1. Out-of-order works only if you're already a senior practitioner filling in specific gaps.
Is there live support or community?
Yes. The bundle includes 6 months in the Engine Room, the practitioner community where Engineering Series students get questions answered. After that, staying is €29 a month or €249 a year.
Where do I get the open-source code (arXiv repos)?
All six repos are linked from the relevant course pages. TrustGate, the security red-teaming kit, the context orchestration code, and the full Enterprise AI OS are Apache 2.0. Links are in the course materials and on the GitHub org.
What if I only want E2 + E5 (the research-anchored ones)?
Buy them separately at €197 each (€394 total) — cheaper than the bundle. You'll miss the assembly story but the two courses stand alone. If you're sure that's all you want, separate purchase is correct.
Time commitment for the full track?
120–180 hours across 6 courses. Self-paced: about 41–52 weeks of evenings and weekends.
Can my company pay for this?
Yes. Invoices issued. Email [email protected] subject 'Reimbursement.'
What's the refund policy?
14-day conditional refund on the €797 bundle. If you have not completed the first module of the first course and you have a real reason to refund, we honor it. After 14 days, no refund.

The complete AI Systems Architect curriculum.

Six courses. 161 lessons. From first API call to deployed platform. €797.

Get on the waitlist
Need just one course? Browse individual courses