4-Week Intensive & 8–12-Week Prompt Engineering Syllabus for Australian VET

A practical prompt engineering syllabus runs from large language model fundamentals to advanced techniques and ends with a hands-on capstone aligned to the national competency NAT11287001. It suits beginners through to working professionals, runs as either an intensive track or a more comprehensive multi-week track, and combines short lectures with labs and a final project.

CanterburyTDI

Build practical AI skills online

Explore CanterburyTDI's nationally recognised, self-paced online vocational courses, including industry-relevant AI training designed to build practical skills.

Explore online courses

Syllabus overview: two practical tracks and module timing
Module-by-module breakdown with lesson topics and lab exercises
Learning outcomes, prerequisites and assessment rubrics
Tools, platforms and practical labs: where to practise and what to monitor
Ethics, prompt injection and responsible AI practice to embed in every module
Suggested capstone projects and portfolio pieces with marking guidance
How this syllabus maps to national training guidance
CTDI study pathway: a practical next step after this syllabus
FAQ
Sources

Syllabus overview: two practical tracks and module timing

A workable prompt engineering curriculum needs more than a list of techniques. It needs a structure that fits around work, study or a career change, with clear hours per module and a sensible mix of theory, lab time and project work.

The shorter intensive track suits learners who already use generative AI tools casually and want to move quickly into applied skills. It compresses fundamentals and core techniques early, then spends the later part on advanced strategies and a condensed capstone. A longer comprehensive track suits complete beginners, career changers or anyone wanting a deeper grounding before tackling domain-specific prompting and deployment basics. It spreads the same content across more weeks, adds extra lab time, and allows for a fuller capstone with milestones and revision cycles.

Both tracks share the same six modules; only the pacing and depth change.

Module 1, LLM fundamentals: prompt anatomy, tokenisation and model behaviour.
Module 2, core prompt techniques: zero-shot, one-shot and few-shot prompting, framing and templates.
Module 3, advanced strategies: role prompts, chain-of-thought reasoning, temperature and prompt tuning.
Module 4, domain prompts: content creation, code and debugging prompts, data extraction.
Module 5, integration and deployment: API basics, context windows, cost and token management.
Module 6, capstone: project brief, milestones, rubric and portfolio guidance.

The syllabus includes six key modules covering fundamentals to capstone project work, with flexible delivery mixing lectures, labs and projects, adapted to course length and depth. Assessment focuses on demonstrating practical prompt engineering skills through outputs such as prompt portfolios, lab exercises, applied projects, prototypes and capstone reviews.

Delivery is typically a mix of recorded or live short lectures, guided labs, and self-directed project blocks, with assessment weighted toward demonstrated output (working prompts, prototypes, a capstone artefact) rather than written exams alone.

Module-by-module breakdown with lesson topics and lab exercises

Each module below pairs a short lesson outline with a lab task a learner can run inside a single sitting.

Module 1, LLM fundamentals. This module covers what a large language model actually does when it receives a prompt: tokenisation, next-token prediction, context windows and the rough shape of model behaviour under different settings. Lessons introduce the anatomy of a prompt, instruction, context, input data, output format and audience, which lines up closely with the framework described in Digital NSW's chatbot prompt essentials guidance for building clear, well-formed prompts.

Lesson 1: how tokenisation works and why it affects cost and output length.
Lesson 2: the five-part prompt structure (instruction, context, input, output format, audience).
Lesson 3: observing model behaviour across temperature and length settings.

Lab: run the same prompt at three different temperature settings, count tokens for each output, and log how wording changes affect length and tone.

Module 2, core prompt techniques. This module builds the fundamentals into working technique: zero-shot prompting (no examples), one-shot prompting (a single example) and few-shot prompting (multiple examples to steer format and tone). Learners practise writing reusable prompt templates and testing variations against each other.

Write a zero-shot version of a task prompt, then a few-shot version with three examples.
Build a reusable template with placeholder fields for a recurring task, such as email drafting or meeting summaries.
Run an A/B test comparing two prompt phrasings on the same task and record which produces more consistent output.

Lab: author three prompt templates for a single use case, A/B test them against each other, and select the best-performing version based on consistency and relevance.

Module 3, advanced strategies. This is where the syllabus moves into role prompting (assigning the model a persona or expertise), chain-of-thought prompting (asking the model to reason step by step) and prompt tuning (iteratively refining a prompt based on output quality). Digital NSW's guidance flags chain-of-thought and prompt tuning as advanced techniques worth dedicated coverage, which this module treats as core content rather than an afterthought.

Lesson 1: role prompts and how persona framing changes tone and depth.
Lesson 2: chain-of-thought prompting for multistep reasoning tasks.
Lesson 3: prompt tuning through iterative refinement and output evaluation.

Lab: design a chain-of-thought prompt for a multi-step problem (such as a basic calculation or logic puzzle), then evaluate three iterations for accuracy and clarity.

Module 4, domain prompts. Learners apply prompting to specific work contexts: content creation (drafting, summarising), code-related prompts (debugging, generating snippets) and data extraction (pulling structured fields from unstructured text). The APS Academy's short practical sessions demonstrate this kind of use-case focus, pairing explanation with live demonstration and hands-on experimentation across drafting and analysis tasks.

Draft a content prompt for summarising a long document into three key points.
Build a debugging prompt that asks the model to identify and explain an error in a short code snippet.
Design a data extraction prompt that pulls names, dates and amounts from a block of unstructured text.

Lab: build a small suite of three to five prompts that together act as a code assistant, covering explanation, debugging and simple refactoring requests.

Module 5, integration and deployment. This module covers the practical side of using prompts in real applications: API basics, managing context windows, tracking token usage and cost, and basic monitoring.

Lesson 1: calling a model API and structuring requests and responses.
Lesson 2: managing context window limits across longer conversations.
Lesson 3: tracking token usage and cost against a simple budget.

Lab: build a small prompt-based application (even a simple script) that logs token counts and cost estimates for each call, creating a basic telemetry record.

Module 6, capstone. The final module sets a project brief, milestone checkpoints, a marking rubric and guidance for turning the finished work into a portfolio piece, covered in full under the capstone section below.

Learning outcomes, prerequisites and assessment rubrics

Each module should map to outcomes a learner can demonstrate, not just describe. Module 1 outcomes centre on explaining tokenisation and prompt anatomy; Module 2 on writing and testing zero, one and few-shot prompts; Module 3 on designing and evaluating chain-of-thought and role-based prompts; Module 4 on applying prompting to a real domain task; Module 5 on building and monitoring a working prompt-based tool; Module 6 on delivering a complete, evaluated capstone project.

Prerequisites are deliberately light. A learner needs basic digital literacy, comfort with web-based tools, and an interest in written or technical communication. Python basics are useful but optional, mainly for Module 5's API work; learners without coding experience can substitute a no-code tool or a spreadsheet-based log for the telemetry exercise.

Rubrics should stay consistent across labs, mini-projects and the capstone, scored against the same core criteria:

Clarity: is the prompt's instruction, context and expected output unambiguous?
Reproducibility: would another person get a similar result running the same prompt?
Safety: does the output avoid fabricated facts, biased framing or unsafe content?
Evaluation: has the learner tested and compared at least two variations before settling on a final version?

Tools, platforms and practical labs: where to practise and what to monitor

Learners can practise most of this syllabus without paid tooling. Free-tier access to a major model playground, combined with a spreadsheet for logging prompts and outputs, covers the first four modules comfortably.

Use a model provider's free playground or trial tier for early experimentation before committing to paid API access.
Keep a simple spreadsheet log of prompt version, output, token count and evaluation notes for every lab.
Use small, anonymised or synthetic test datasets rather than real personal or client data when practising extraction or classification prompts.
Track token usage per task to build a feel for cost before moving to production-style deployment.

For Module 5's deployment work, a small budget cap (even a few dollars of API credit) is usually enough to complete the telemetry lab. Learners validating AI-generated outputs more rigorously, particularly for code tasks, can draw on broader software testing practices for generative AI as a reference for structuring validation checks.

Pro Tip: Log every prompt variation you test, even failed ones. The failed attempts often teach you more about model behaviour than the one that worked.

Ethics, prompt injection and responsible AI practice to embed in every module

Responsible use should run through every module rather than sit in a single ethics unit at the end. Prompt injection, where hidden or malicious instructions embedded in input text try to override a prompt's original intent, is worth demonstrating early with a simple classroom exercise: give learners a prompt with a deliberately embedded conflicting instruction and ask them to spot it before running it.

Conflicting prompt instruction stopped before execution

Bias and hallucination checks fit naturally into Modules 2 through 4: after generating an output, learners compare it against a known-correct answer or a second model run to spot inconsistencies. GovAI's interactive learning resources frame this kind of safety and ethics guidance as something to teach alongside technique, not as a separate add-on.

A short classroom policy covering three points keeps assessment integrity intact:

No real personal, client or commercially sensitive data in any lab or capstone submission.
All AI-generated work submitted for assessment must be disclosed as such.
Learners must manually verify any factual claim an output makes before including it in a deliverable.

One practical safeguard to teach early: the National AI Centre's practical training resources include responsible AI topics as core material in free, self-paced microskills, reinforcing that safety checks belong in week one, not week twelve.

Suggested capstone projects and portfolio pieces with marking guidance

A good capstone gives learners a real deliverable they can show a future employer or client, not just a grade.

Multi-role study assistant. A prompt suite that adapts explanations to different learner levels (beginner, intermediate, advanced) on a chosen topic, demonstrating role prompting and few-shot framing.
Code helper. A small prompt suite that explains errors, suggests fixes and writes basic unit tests for a given code snippet, demonstrating domain-specific prompting from Module 4.
Summariser with extractive question answering. A tool that takes a long document, produces a summary and answers specific factual questions by extracting (not inventing) the relevant passage.
Domain-specific education chatbot. A prompt-based assistant scoped to a single subject area, with guardrails against answering outside its scope and a short test log showing how it handles edge cases.

Deliverables should include the working prompt set, a short demo (recorded or live), a one-page write-up explaining design choices, and the evaluation log showing at least two rounds of refinement. Marking checkpoints typically fall at the brief submission, a midpoint milestone review, and the final demo, scored against the clarity, reproducibility, safety and evaluation criteria used throughout the course.

When publishing a capstone as a portfolio piece, learners should strip any real personal or organisational data, use synthetic examples where the original task involved sensitive content, and note clearly that the piece was built for a learning exercise rather than a production deployment.

How this syllabus maps to national training guidance

This structure draws directly on NAT11287001, the national unit that formalises prompt engineering competencies for generative AI in Australia, and on free practical microskills from TAFE NSW and the National AI Centre covering AI foundations, generative AI applications and responsible AI practice. Mapping module outcomes back to a recognised unit code makes it far easier to adapt this syllabus into a short course, a workplace training session, or formal VET unit delivery, rather than treating prompt skills as a loose set of tips.

Educators adapting this outline for formal delivery should keep the module structure intact while adjusting hours and assessment weighting to match their own unit requirements, and should treat the capstone as the primary evidence of competency rather than quizzes alone.

CTDI study pathway: a practical next step after this syllabus

If this syllabus has you thinking about a formal qualification rather than a self-directed course, our Diploma of Artificial Intelligence covers much of the same practical ground, applied prompt design, project work and hands-on labs, through flexible, self-paced online delivery, designed and taught by industry experts.

CanterburyTDI

We built the diploma around the same principle this syllabus follows: skills demonstrated through real project work, not just theory. Studying online means you can work through modules around a job, caring responsibilities or other study, at your own pace, with support from our team along the way.

Nationally accredited qualification delivered online.
Course content designed with a focus on practical, employment-relevant skills.
Self-paced structure suited to working professionals and career changers.

We do not guarantee job outcomes, salary increases or credit transfer into other programs, those depend on individual circumstances and the policies of other institutions. If you want to see how the diploma is structured, view course details and enrol to check current intake dates and the one-off enrolment fee.

FAQ

What are the topics in prompt engineering?

Prompt engineering topics typically span LLM fundamentals and tokenisation, zero-shot, one-shot and few-shot prompting, advanced strategies like chain-of-thought and role prompts, domain-specific prompting for content, code and data tasks, plus deployment basics and responsible AI practice. The national unit NAT11287001 groups these into formal competency statements for Australian training purposes.

What topics are covered in a prompt engineering course?

A structured course generally covers the same ground as the syllabus above: foundations and prompt anatomy, core prompting techniques, advanced strategies such as prompt tuning, domain applications, integration and deployment, and a capstone project. Courses vary in depth and pacing, from short microskills to multi-week programs.

Is prompt engineering difficult to learn?

Prompt engineering is accessible to most adult learners with basic digital literacy, since the core techniques (framing, examples, structure) do not require coding. It becomes more demanding at the advanced end, where chain-of-thought design, prompt tuning and deployment work benefit from some technical comfort, though this can be built progressively through labs rather than needed from day one.

How do you learn prompt engineering?

Most learners progress by combining short lessons with hands-on labs: writing and testing prompts, comparing variations, and applying techniques to real tasks like drafting, summarising or code debugging. Free resources such as the National AI Centre's practical training hub and structured courses both support this kind of applied, project-based learning.