Skip to main content

The Roles This Book Trains

The market is inventing titles faster than it can define them. Most of those titles are the same discipline at different depths: the discipline this book teaches. Here is the map, and exactly how far the book carries you toward each role.

The titles below define this new vocabulary, sorted by how the work actually clusters. The verdict beside each role is the honest scope line: how far this book carries you toward it, and where the certification tracks take over. The verdicts matter more than the names. Where the book stops, it says so.

New to the vocabulary (Digital FTE, SKILL.md, Agent Factory)? Start with the Thesis and the Glossary; this page assumes them.

The generalist core​

AI Engineer: the generalist builder. The market's current catch-all: someone who builds applications out of AI components and drives AI coding agents. This is the book's primary graduate. Spec-driven development, SKILL.md authoring, agent architecture, tool and MCP interfaces, evaluation, cloud-native deployment, human oversight. The full spine. Trains it, end to end.

Agent Factory Architect: the builder who designs the line, not the part. Not one agent: the whole manufacturing process by which Digital FTEs are specified, built, deployed, and governed. The book is its canonical source; the five-quarter Certified Agentic AI Architect program is its credential. Trained in full; certified by the Architect track.

The two roles only this book trains​

Subject Matter Expert as Skill Author: the role the market hasn't named yet. The accountant, lawyer, or supply-chain expert who encodes judgment into SKILL.md and becomes the knowledge engine of a Digital FTE. Most market lists miss this role because they still picture AI work as engineering-only; this book treats domain judgment itself as something to be authored, tested, and deployed. It is one of two roles only this book trains. Trains it in full: judgment in, working agent out.

Forward Deployed Engineer (FDE): the vendor-neutral version the market can't find. The FDE (pioneered by Palantir, now revived by AI vendors embedding engineers inside client organizations) customizes agentic workflows to a client's reality. That technical half is Agent Factory work, and the book trains it fully. The twist is vendor neutrality. Andrew Ng has noted in The Batch that clients struggle to find FDEs who aren't tied to a single vendor: the role exists to wire one vendor's product deep into the company, which erodes the freedom to switch later. The method here is bound to none, so the book produces the FDE the market can't source, with that optionality intact. That is the second role only this book trains. The other half of the job (client discovery, prioritization, ROI framing, the discipline to push back on an unrealistic ask) belongs to the Certified Agentic AI Business Strategist track, not the core book. Trains the technical core; the consulting layer lives in the Strategist track.

The supporting roles​

Evals Engineer: the verification specialist. Verification is not an afterthought here; it is a survival standard. Core curriculum, not an add-on.

Digital FTE Supervisor: the governance role. The human who owns accountability for a deployed AI worker: the human-in-the-loop, the reviewer, the name the audit trail points to. It is the operator's job (running the work, not building the worker) turned into a title. Trains it.

Where the book deliberately stops​

LLMOps Engineer: depends what you mean. Running agents reliably in production (orchestration, observability, scaling on Dapr and Kubernetes) the book trains. Training, serving, or fine-tuning the underlying models: out of scope, by design. The book sits at the application and discipline layer above the model, where the work survives the model commoditizing. Trains the production-ops side; model internals are deliberately not ours.

Harness Engineer: build the one around your Worker, not the one you work inside. The word covers two different things, which is what makes this look like a contradiction. You do build and deploy the control-plane harness that runs your own Worker in production: the layer that holds secrets, runs the agent loop, keeps state, and stays split from the sandbox where generated code executes. That is core deployment work the book trains in full, not something it stops short of. What it leaves alone is building a general-purpose agent tool (Claude Code, OpenCode) for other people to operate inside. The thesis is that your discipline outlives whichever of those tools wins, so you operate across them rather than reinventing one. Trains the harness you deploy around your Worker; not the general-purpose agent tool you run inside.

AI Data Engineer: the agent-facing data layer. The system-of-record work (Postgres, pgvector, MCP as the spine an agent reads from) touches the agent-facing data layer. Classic pipeline and warehouse engineering is adjacent, not central. Trains the agent-facing data layer, not general data engineering.

The pattern is the tell. Every role the market is naming is a builder of tools for the agent era. The two this book uniquely trains (the expert who authors skills, and the engineer whose method is bound to no vendor) are the two the market can't yet manufacture and most needs.