New · AI Testing Engineer

AI Testing Engineer — Evaluation & Governance

Testing generative AI is its own craft. This track teaches the three skills that separate an AI testing engineer from a tester who happens to use AI.

You already know why AI testing differs from traditional testing, and how to test data, models, and fairness against the standard. This track goes a layer deeper into the systems teams actually ship today: retrieval-augmented generation, prompts that can be attacked, and agents that take actions on their own. Each lesson is hands-on, with three prompt-lab exercises that mark your work against a real LLM.

This track covers

RAG Evaluation Prompt-Injection Testing Agent Testing Faithfulness & Grounding Guardrails Human-in-the-Loop Metamorphic Testing Neural Network Coverage

Builds on

The CT-GenAI foundations and the ISO/IEC 42119 module. This track assumes you understand what a large language model is, the five AI failure modes, and risk-based AI test scope.

Who this is for

Senior testers, Test Leads, and QA engineers putting generative AI features into NZ production systems — chatbots, document assistants, and agents that act on a customer's behalf.

The 8 lessons

Evaluate. Attack. Govern.

Lesson 1

RAG Evaluation

Evaluating retrieval-augmented generation. Faithfulness, answer relevance, and context precision and recall. How to build a RAG eval set, and the confident-but-ungrounded failure that passes every smoke test.

~30 min read · ~70 min with exercises · AI Evaluation

Lesson 2

Prompt-Injection Testing

Defensive prompt-injection vulnerability testing. Direct and indirect injection, jailbreaks, and data exfiltration. Testing system-prompt robustness and writing the defensive test cases that catch it before an attacker does.

~30 min read · ~75 min with exercises · AI Evaluation

Lesson 3

Agent Testing

Testing AI agents and agentic systems. Multi-step tool-use validation, non-determinism, guardrails, and human-in-the-loop sign-off. Model benchmarking and deterministic-consistency checks for systems that take actions.

~35 min read · ~80 min with exercises · AI Evaluation

Lesson 4

Model Benchmarking

Benchmarking model quality the right way: task-specific eval sets, golden datasets and scoring rubrics, A/B comparison of models and prompts, accuracy-vs-latency-vs-cost trade-offs, and regression-gating a model upgrade before it ships.

~30 min read · ~75 min with exercises · AI Evaluation

Lesson 5

Deterministic & Consistency Testing

Testing non-deterministic output: temperature and seed control, semantic-equivalence assertions instead of exact-match, tolerance bands, repeat-N stability checks, and the flaky-assertion patterns that wreck a generative test suite.

~30 min read · ~75 min with exercises · AI Evaluation

Lesson 6

Human-in-the-Loop Sign-off

Governance sign-off frameworks: when a human must approve, confidence thresholds and escalation, audit trails, reviewer sampling, four-eyes approval, and a RACI for AI decisions under the Privacy Act 2020 and public-sector accountability.

~30 min read · ~75 min with exercises · AI Evaluation

Lesson 7

Metamorphic Testing Relations

Testing AI systems when no oracle exists. The four MR types — invariance, monotonicity, equivalence, consistency — how to spot spurious correlations like demographic bias, and how to assert MRs on probabilistic LLM output.

~30 min read · ~75 min with exercises · CT-AI v2.0

Lesson 8

Neural Network Coverage Metrics

Why 99% accuracy is not enough. Neuron Coverage, k-multisection (k-MNAC), Neuron Boundary Coverage, SNAC, and layer-wise analysis — how to identify which parts of a model your test suite has never exercised.

~30 min read · ~75 min with exercises · CT-AI v2.0

Testing approaches compared

Different parts of an AI system need different testing approaches. Use this to decide which combination applies to your system.

Approach What it verifies What it misses Automation When to use
Unit / function testIndividual helper functions, prompt builders, parsersModel behaviour, reasoning qualityFullAlways — fast, cheap, catches regressions in glue code
Prompt regression testThat a known prompt still produces an acceptable outputNovel inputs, distribution shiftFullBefore every model or prompt change
API / integration testService behaviour, contract conformance, latency SLAsAgent decision quality, multi-step reasoningFullWhenever the AI system exposes an API surface
Agent / end-to-end testMulti-step task completion, tool use, goal achievementReasoning quality, edge cases, adversarial inputsPartialFor any agent with real-world tool access or user impact
LLM-as-judge evalOutput quality at scale — relevance, faithfulness, toneReproducibility — judge model introduces its own varianceFullWhen human review cannot scale to cover the eval dataset
Security / red teamPrompt injection, jailbreaks, data extraction, IDORNovel attack patterns not in the scanner libraryPartialMandatory before any AI system with external-facing access
Human reviewContext, judgement, cultural nuance, ethical concernsScale — cannot review every output in productionManualHigh-stakes decisions, regulatory sign-off, tone/cultural checks
Why this track

The systems you ship now are generative

Most AI features going live in NZ today are not custom-trained models. They are generative AI wired into a business: a retrieval system that answers customer questions from policy documents, a prompt that summarises a case file, an agent that books an appointment or files a form. These systems fail in ways a model-accuracy test never catches. They answer confidently from the wrong document. They follow instructions hidden in the data they read. They take an action no one signed off on.

This track teaches you to test for exactly those failures. Lesson 1 makes the output measurable — is the answer actually grounded in what the system retrieved? Lesson 2 takes the attacker's view defensively — what happens when the input is hostile? Lesson 3 governs the riskiest case of all — an AI that acts, not just answers.

Throughout, the context is NZ: a HealthNZ patient assistant, an Revenue NZ help bot, an Benefits NZ case agent, RealMe identity flows, and the Privacy Act 2020. If an AI feature shapes a decision about a person, you will be asked how it was tested. This track is how you answer.

Context

How we got here

AI testing techniques evolve in response to what teams actually ship. Each wave of system architecture introduced failure modes the previous testing approach could not catch.

2017–2020 — Traditional ML

Testing focused on accuracy, precision, recall, and AUC on static datasets. Inputs were structured; outputs were classifications. Exact-match assertions worked. Fairness was rarely measured systematically.

2021–2022 — Foundation models go mainstream

Outputs became free text. The same prompt produced different wording each run. Exact-match assertions became flaky by design. Teams needed semantic-equivalence assertions, repeat-N stability checks, and new tools for asserting on meaning rather than strings.

2023 — RAG patterns gain adoption

Organisations wired LLMs to their own knowledge bases. Retrieval quality, faithfulness, and document staleness became primary failure modes. Teams learned that retrieval recall and answer faithfulness are different metrics requiring different tests.

2024 — Agentic AI emerges

AI systems began taking actions: booking appointments, sending emails, filing forms, initiating payments. Wrong answers became wrong actions. Testing had to cover action safety, tool-use validation, unbounded execution loops, and human approval workflows.

2025 — Governance frameworks mature

ISO/IEC 42119, CT-AI v2.0, and the OWASP LLM Top 10 gave structure to what teams were discovering empirically. Metamorphic testing for fairness, neural network coverage, and prompt injection testing moved from research into standard practice for regulated systems.

2026 — Operational assurance

The focus shifts from “can the model do this?” to “can we trust it to keep doing it in production, at scale, under adversarial conditions, across demographic groups, with full audit trails?” This track teaches those skills.

Decision guide

Which technique, and when?

Match your situation to the technique. Most real systems need more than one.

Your situationStart with
Output varies on the same input — test suite is flakyDeterministic-Consistency Testing
No correct answer to assert against (no test oracle)Metamorphic Testing
System answers from retrieved documentsRAG Evaluation
Users can type free-form input to the AIPrompt Injection Testing (mandatory)
AI takes actions — pays, books, files, sendsAgent Testing + HITL Sign-Off
Comparing two models or prompt versionsModel Benchmarking
Decision affects individuals — needs human accountabilityHuman-in-the-Loop Sign-Off
Need to verify fairness across demographic groupsMetamorphic Testing (demographic MRs)
Test suite may have blind spots in model behaviourNeural Network Coverage
Related

Where this fits