Resources › Professional development
AI-Augmented Systems Engineering
A professional-development course on systems engineering for a new class of system — one where artificial intelligence is both a tool the engineer uses and a component the engineer must build. Every slide deck is free to download below.
What the course covers
The course is grounded in International Council on Systems Engineering (INCOSE) practice and assumes a first course in systems engineering. It applies that foundation to systems that contain a learned, probabilistic component, where the classical assumption of deterministic parts no longer holds. It is written for engineers across every discipline: each team carries one system from its own field — a delivery drone, an infusion pump, a smart bridge, an electric-vehicle battery pack, a robotic warehouse — through the full lifecycle.
The material is organized around two dualities:
- AI for systems engineering
- AI accelerates the engineer’s own work — requirements analysis, design-space exploration, verification, and documentation done faster and at larger scale. The engineer stays the originator, the judge of quality, and the resolver of the tool’s failures.
- Systems engineering for AI
- A learned element behaves probabilistically, depends on its training distribution, and resists inspection. It must be specified, allocated, verified, assured, and certified like any other part — but by new means.
Why this matters now
For most of engineering’s history, most engineers worked as implementors. Given a specification, they produced the artifact — the code, the register-transfer-level (RTL) description, the drawing, the analysis. A smaller group of systems engineers worked at the level of the whole: requirements, interfaces, function allocation, integration, verification, and the trade-offs among them.
AI collapses the cost of implementation. When a model can draft the code, the test bench, the document, or the analysis in seconds, producing the artifact is no longer the scarce, decisive work. What stays scarce is everything above it — deciding what to build, specifying it precisely enough to be checkable, defining the interfaces between many fast-generated parts, verifying that the result meets the need, and owning the outcome. That work is systems engineering.
So AI does not only make implementors faster; it raises their altitude. Any engineer who uses AI to generate work is pushed, at least in part, into the systems engineer’s role. The bottleneck moves from “can I build this piece” to “can I specify, integrate, and assure the whole.” As output grows, so does complexity — more components, more interfaces, more integration points, faster change. Managing that complexity, through boundaries, interfaces, requirements traceability, verification, and risk, is exactly the systems engineer’s craft. This course teaches that craft deliberately, and treats AI itself as one more component that must be specified, verified, and signed for.
How it is organized
The material is organized in three parts — foundations, assurance, and synthesis — and follows a single running example: one system carried from its concept of operations and requirements, through architecture, modeling, and a trade study, into verification, reliability, risk, certification, and the human role. Each lecture is a self-contained slide deck.
Slide decks
The decks are Microsoft PowerPoint files (.pptx). Most are 110–150 KB; the overview and the expanded Week 1 deck are larger. Start with the overview, which sets up the framing and the roadmap.
Download the course overview deck
Part 1 · Foundations
The lifecycle and analysis toolkit: how the work is structured, what a requirement is, where function is allocated, and how a system is modeled and searched.
| # | Lecture | Focus |
|---|---|---|
| 1 | The AI-Augmented Systems Problem | Two dualities: AI as a tool for doing engineering, and AI as a component the engineer must engineer. |
| 2 | Systems Thinking & Lifecycle Models | The Vee and ISO/IEC/IEEE 15288 lifecycle processes, and how AI-in-the-loop workflows compress and stress the Vee. |
| 3 | Stakeholder Needs, ConOps & Requirements Engineering | Elicitation, the concept of operations, and verifiable requirements with traceability; specifying a learned component. |
| 4 | Architecture & Function Allocation | The central act of allocating function among human, classical-algorithmic, and AI/ML elements. |
| 5 | Model-Based Systems Engineering & Systems-as-Graphs | Model-Based Systems Engineering (MBSE) and SysML, the digital thread, and representing learned components with their evidence. |
| 6 | Design-Space Exploration & Trade Studies | Multi-objective optimization, Pareto analysis, and weighted trade studies, with AI as an explorer of large design spaces. |
| 7 | Graph-Theoretic Foundations for Systems Analysis | Systems as graphs: design-structure and N-squared analysis, partitioning, critical paths, and graph neural networks. |
Part 2 · Assurance
Making the case that the system works: verification and validation of learned components, reliability, risk, program control, and certification.
| # | Lecture | Focus |
|---|---|---|
| 8 | Verification & Validation of Systems with Learned Components | V&V planning, statistical testing of learned components across their operating conditions, and assurance cases. |
| 9 | Reliability, Availability & Resilience (RAMS) | Reliability, availability, and maintainability; fault tolerance and graceful degradation with AI components in the loop. |
| 10 | Integrated Risk Management | Risk identification, analysis, and handling across the lifecycle; AI-specific technical risks and programmatic risk. |
| 11 | SE Management: Cost, Schedule & Technical Performance | Work breakdown structures, earned value management, technical performance measures, and design reviews. |
| 12 | Certification, Assurance & Accountability | Why certification regimes exist, the evidence they demand across industries, and accountable human sign-off. |
Part 3 · Synthesis
The human role and the close: where judgment stays decisive, how integration and agentic workflows change the work, and the framework that ties the course together.
| # | Lecture | Focus |
|---|---|---|
| 13 | Human-Systems Integration & the Limits of Automation | Human-systems integration, tacit knowledge, and the deskilling and pipeline risk, with countermeasures. |
| 14 | Systems Integration & Test; Agentic Futures | Integration strategy and interface management, agentic AI workflows, and the limits that persist. |
| 15 | Capstone Presentations & Course Synthesis | Team capstone presentations and the complementarity framework as the course’s closing lens. |
The complementarity framework
The course closes on a single lens for placing AI in an engineering system. AI plays three roles, and each leaves a distinct human responsibility.
- Accelerator
- AI does faster what engineers already do. The human stays the originator, the judge, and the one who fixes what the tool gets wrong.
- Explorer
- AI searches spaces too large to enumerate. The human sets the space and picks the operating point.
- Pattern recognizer
- AI sees signatures below human perception. The human owns the detection design and the response.
Allocate by mode, verify by evidence, certify by accountability, and engineer the human role deliberately.
Where this fits in a curriculum
These lectures form an overlay on a standard graduate systems-engineering curriculum: the systems-engineering method is the foundation, and the treatment of AI as both tool and component is what this course adds. For a worked example, see how the lectures map onto a graduate systems-engineering program.
Use and licensing
These materials are published by the FPGA Professional Association for professional development. Written content on this site is available under CC BY 4.0. To ask about running the course or contributing improvements, email contact@fpgapa.org.