Architect Role & Mindset

What this guide is

A guide to the architect’s job, not about system design.

That distinction decides everything else. It does not teach distributed systems patterns, database selection or deployment engineering — it assumes you know them. It teaches what an architect actually does all day: read business context, turn it into requirements, choose and justify a technical strategy, earn the trust that makes it stick, run a process that outlives you, and evolve systems inside organisations that resist.

Ten modules in four parts, in two halves: a Handbook to read, and an optional Workbook of ten exercises.

Who it is for

An experienced engineer or tech lead moving toward a solution architect role, studying alone, without a cohort or an instructor.

Assumed prior knowledge — this guide does not teach it

You are expected to arrive knowing: distributed systems patterns and their failure modes · CAP and consistency models · database, queue, cache and load-balancer taxonomies and selection criteria · Kafka · caching strategies · OAuth 2.0 / OIDC / JWT / RBAC / ABAC · microservice mechanics (saga, circuit breaker, service mesh, API gateway) · basic RAG and vector-store mechanics · general systems thinking.

Where a module needs one of these, it references it in a line and moves on. If several are unfamiliar, a technical-leadership or distributed-systems course is the better starting point.

Explicitly out of scope

Hands-on coding · model training and fine-tuning implementation · cloud-provider certification material · people management, meaning hiring, firing, one-on-ones, OKRs and budgets.

The line this guide holds: it teaches you to influence the organisation, not to own it.

How it is built

One running case across the whole guide. Every module reasons about a single fictional but fixed company, Lintel — a B2B field-service business with a legacy monolith, no spare engineering capacity, and an AI support initiative to deliver. In the Workbook you design that system yourself, one exercise per module, and the last exercise assembles everything into an architecture portfolio for a realistic engagement.

Every module uses the same fixed facts, and you can substitute your own system if it has a genuine business driver with a number attached, conflicting stakeholders, a legacy component you cannot replace, a compliance constraint, and a plausible AI component.

Part 1 — UnderstandModules 1–2. What the architect's job is, and how a business need becomes a requirement that shapes structure.

Part 2 — DecideModules 3–5. Choosing a structure, the AI stack, and how much to delegate to agents.

Part 3 — Build & OperateModules 6–7. Running it in production — cost, safety, evaluation — and designing under uncertainty.

Part 4 — SustainModules 8–10. Working through the organisation, making the practice repeatable, and evolving what exists.

The Workbook — optionalTen exercises, one per module, each consuming what the previous ones produced. Done in order, they build one portfolio.

The Lintel caseThe fixed facts, constraints and stakeholder cast every module draws on. Read this before Module 1.

How to work it

The Handbook is read in order — each module builds on the ones before it, and each topic ends in questions worth answering rather than skimming. The AI modules (4–7) are self-contained enough to read out of sequence if that is all you came for.

The Workbook is optional, and it is where the learning actually happens: reading gives you recognition, practice gives you capability. Do an exercise after reading its Handbook module — or work through all ten once you have finished the Handbook. Either way, work them in order or not at all: every exercise consumes artifacts the earlier ones produced, so a skipped exercise leaves a later one with nothing to build on.

The honest limit, stated once: the guide can give you the frameworks, the arithmetic habits and a complete worked engagement. It cannot give you feedback from production, which is the only thing that confirms a judgement was right. Module 10 returns to that.