---
title: "AI governance engineer: what they do, skills, salary"
description: "What an AI governance engineer does: the seven workflows the role owns, the skills behind them, IAPP salary medians and three ways into the role."
canonical: https://aigovernanceengineer.com/role
author: "Jorge García Aibar"
license: "CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)"
doi: https://doi.org/10.5281/zenodo.22956197
version: "0.5.0"
updated: 2026-09-26
---

# AI governance engineer: what they do, skills, salary

> The person, on whatever org chart, who holds the capability of AI governance engineering and is accountable for the three questions in production: what AI is running, what it is allowed to do, and what evidence proves it.

## In short

An AI governance engineer holds the capability of AI governance engineering and is accountable for three questions in production: what AI is running, what it is allowed to do, and what evidence proves it. The role owns seven workflows, each a running system rather than a document: intake and classification, inventory and registry, evals and red teaming as evidence, policy-as-code and gates, runtime monitoring and incidents, assurance and audit evidence, and regulatory translation. The load-bearing skills follow them: eval harness engineering, policy-as-code in OPA/Rego or Cedar, CI/CD pipelines, observability and incident response, OSCAL evidence, and enough law-reading to parse an article without mistaking it for advice. The analyst describes the system from the outside; the engineer reads it directly and ships the control. People come in from legal or privacy, from security or GRC, and from MLOps. The IAPP Salary & Jobs Report 2025-26 puts the median technical AI-governance salary at USD 221k in the tech sector.

## What does an AI governance engineer do?

An AI governance engineer owns seven workflows, not documents: intake and classification, inventory and registry, evals and red teaming, policy-as-code and gates, runtime monitoring and incidents, assurance and audit evidence, and regulatory translation. Each is a running system with inputs, outputs and the stack layer it maps to.

- **Intake and classification** (Layer 2 (Inventory & Transparency)): Every AI system, model and agent enters through an intake that classifies it by risk tier, regulatory exposure, data sensitivity and autonomy. The engineer builds intake as a form-plus-code path that scaffolds a registry entry, triggers the right impact assessment and routes the system to the controls its class requires. [Chapter 06](https://aigovernanceengineer.com/bok/the-role#intake-and-classification)
- **Inventory and registry** (Layer 2 (Inventory & Transparency)): The engineer owns the inventory of models and the agent registry: every non-human actor with an owner, a declared scope, a status and a kill switch. What distinguishes the engineer is the runtime data path: the registry is fed by the deployment pipeline and by discovery against production, not typed into a spreadsheet. [Chapter 06](https://aigovernanceengineer.com/bok/the-role#inventory-and-registry)
- **Evals and red teaming as evidence** (Layer 3 (Evals & Red Teaming as Evidence)): The engineer builds and maintains the eval suites (capability, safety and adversarial) and wires them into an eval gate so a failing eval blocks the release. This is the workflow that most sharply separates the engineer from the analyst: the engineer writes the test the model must pass and owns the harness that runs it. [Chapter 06](https://aigovernanceengineer.com/bok/the-role#evals-and-red-teaming-as-evidence)
- **Policy-as-code and gates** (Layer 1 (Govern-as-Code)): The engineer expresses governance rules as executable policy that evaluates in CI/CD and at admission, and maintains the gates that enforce them. The output is a merge that is blocked or allowed, with a logged reason, not a recommendation in a review. [Chapter 06](https://aigovernanceengineer.com/bok/the-role#policy-as-code-and-gates)
- **Runtime monitoring and incidents** (Layer 4 (Runtime Controls & Observability)): The engineer instruments runtime: guardrail decisions, tool-call mediation, drift signals and agent behaviour stream into observability. They own the detection-to-report path for serious incidents, including the Article 73 clock for high-risk systems, and the tested kill switch for agents. [Chapter 06](https://aigovernanceengineer.com/bok/the-role#runtime-monitoring-and-incidents)
- **Assurance and audit evidence** (Layer 5 (Assurance & Continuous Compliance)): The engineer emits audit-ready evidence as a by-product of the build (OSCAL component and assessment artefacts, signed logs, structured eval results), so the audit is a query, not a project. This is continuous assurance: the control status is a live signal, not a point-in-time attestation. [Chapter 06](https://aigovernanceengineer.com/bok/the-role#assurance-and-audit-evidence)
- **Regulatory translation** (all layers): The engineer reads the obligation well enough to build the control that meets it, turning an AI Act article, an ISO/IEC 42001 control or a NIST AI RMF subcategory into a gate, a registry field or an evidence artefact, and back, so an auditor can trace the control to the obligation. This is translation, not legal advice. [Chapter 06](https://aigovernanceengineer.com/bok/the-role#regulatory-translation)

## What skills does an AI governance engineer need?

The skills follow the workflows. The load-bearing ones are eval harness engineering, policy-as-code in OPA/Rego or Cedar, CI/CD pipeline engineering, observability and incident response, OSCAL and machine-readable evidence, and reading the AI Act, ISO/IEC 42001 and the NIST AI RMF. Under all seven: enough Python to glue systems together.

| Workflow | Core skills | Supporting skills |
|---|---|---|
| Intake & classification | Risk taxonomy design; Reading the EU AI Act risk tiers; Requirements analysis | Form/workflow tooling; Light data modelling |
| Inventory & registry | Non-human identity and scoped access; API integration to CI/CD; Data modelling | Cloud IAM; Discovery tooling; SPIFFE/SPIRE concepts |
| Evals & red teaming | Eval harness engineering; Adversarial prompting; Statistical literacy; Python | LLM/agent internals; Benchmark design; Threat modelling (STRIDE/PASTA) |
| Policy-as-code & gates | OPA/Rego or Cedar; CI/CD pipeline engineering; Git | Policy schema design (Policy Cards); Admission control |
| Runtime monitoring & incidents | Observability/OpenTelemetry; Guardrail configuration; Incident response | Detection engineering; MCP and agent-protocol security |
| Assurance & audit evidence | OSCAL and machine-readable evidence; Logging and signing; Audit fluency | Cryptographic attestation; Evidence-store design |
| Regulatory translation | Reading regulation and standards (AI Act, ISO/IEC 42001, NIST AI RMF); Mapping | Legal English; DPIA/FRIA methodology |
| Cross-cutting (under all seven workflows) | Python: enough to glue systems together; Law-reading: enough to parse an article without mistaking it for advice | None |

Read in chapter 06: https://aigovernanceengineer.com/bok/the-role#skills-by-workflow

## One line separates the two

The AI governance analyst describes the system from the outside and files the description; the AI governance engineer reads the system directly and ships the control that changes what it does.

| Dimension | Analyst | Engineer |
|---|---|---|
| Evidence artefact | A point-in-time artefact (an attestation, a questionnaire, an exported report) compiled for a review | A continuously emitted artefact (a query result, an eval run, a signed log) produced as the pipeline runs |
| Primary source | Works from the system's reported description: documentation, summaries and vendor answers | Works from the running system: the registry and production telemetry, the same signals the build emits |
| Toolset | Spreadsheets, a GRC/AI-governance platform, ticketing | Python, OPA/Rego, Git, CI/CD, eval harnesses, OSCAL, plus the platform |
| Cadence | Periodic: quarterly reviews, annual assessments | Continuous: every commit, deploy and runtime call |
| Output | A report, a mapping matrix, a risk rating | A merged-or-blocked build, a registered agent, a machine-readable evidence artefact |
| Success metric | Audit passed, framework coverage complete | Control failures caught before the auditor arrives, and realised risk measurably reduced |

## How do you become an AI governance engineer?

There are three ways in, each from an adjacent discipline, and each brings an edge and a gap. From legal or privacy, start at policy-as-code and intake. From security or GRC, start at evals-as-evidence and the agent registry. From MLOps or ML engineering, start at eval gates in CI and runtime monitoring.

- **From Legal or privacy**: Your edge is regulatory translation; your gap is the build. Turn one assessment into a versioned, executable artefact and learn enough pipeline to see where the control fires. Start at policy-as-code and intake.
- **From Security or GRC**: Your edge is the control mindset; your gap is the model layer. Add the AI-specific objects (evals as controls, agent identity and scope, and the model and agent failure modes of the OWASP Agentic Top 10), starting at evals-as-evidence and the agent registry.
- **From MLOps or ML engineering**: Your edge is the runtime data path everyone else lacks; your gap is the obligation. Add the eval gate rather than the eval report, the registry field for owner and scope, and the evidence artefact the audit needs. Start at eval gates in CI and runtime monitoring.

Read in chapter 06: https://aigovernanceengineer.com/bok/the-role#three-ways-in

## Next step: start one workflow this week

Whichever way in, a first hire owns two or three workflows. Each tool below starts one of them in the browser and leaves you a record to keep.

- [Classify one system with the EU AI Act risk classification checker](https://aigovernanceengineer.com/toolkit/ai-act-triage): Intake and classification: indicative roles and risk classes, each with its article, and a decision record to file.
- [Start the registry with the AI register entry builder](https://aigovernanceengineer.com/toolkit/ai-register-entry): Inventory and registry: entries with an owner, a scope and an expiry that validate against the published schemas.
- [Turn one rule into policy-as-code with the Policy Card builder](https://aigovernanceengineer.com/toolkit/policy-card): Policy-as-code and gates: a Policy Card, an OPA/Rego module with unit tests and the CI hook that runs it.

Indicative, not legal advice and not a conformity claim. Nothing you enter leaves your browser.

## Three tells an employer has the role wrong

1. Certifications as a proxy for capability: descriptions list AIGP, CIPP, CISSP and CISM as if a certificate produced a control, while under-specifying the load-bearing skills: eval harnesses, policy-as-code, the runtime data path.
2. Analyst work under an engineer title: in our reading of the postings, "AI Governance Engineer" titles often describe intake, mapping and reporting at engineer pay; the tell is the absence of any build: no eval gate, no registry integration, no evidence pipeline.
3. Everything, in one hire: a single posting asks for policy-as-code, red teaming, identity, observability, incident response, regulatory translation and stakeholder management. That is a function, not a person; a first hire owns two or three workflows and builds the paved path for the rest.

## How much does an AI governance engineer earn?

The IAPP Salary & Jobs Report 2025-26 puts the median technical AI-governance salary at USD 221k in the tech sector, its highest band. The median across AI-governance work generally is USD 151.8k, and USD 169.7k for combined privacy-and-AI-governance roles. These are primary medians, not job-board averages.

- USD 221k: Median technical AI-governance salary, tech sector (highest band). Source: [IAPP Salary & Jobs Report 2025-26](https://iapp.org/resources/article/salary-survey-summary/)
- USD 151.8k: Median AI-governance salary (work generally). Source: [IAPP Salary & Jobs Report 2025-26](https://iapp.org/resources/article/salary-survey-summary/)
- USD 169.7k: Median combined privacy-and-AI-governance salary. Source: [IAPP Salary & Jobs Report 2025-26](https://iapp.org/resources/article/salary-survey-summary/)

## Maturity, from paper to production, in five levels

- Level 1, Documented: Governance exists as artefacts a human maintains: a policy PDF, a spreadsheet inventory, a risk register, a review before launch. The rules are written and someone is accountable, but nothing executes. [Chapter 07](https://aigovernanceengineer.com/bok/maturity-model#the-five-levels)
- Level 2, Inventoried: There is a real inventory of models and an agent registry, fed by a runtime data path rather than typed by hand: a deploy registers a system with an owner, a scope and a status. You can answer what is running on any given day. [Chapter 07](https://aigovernanceengineer.com/bok/maturity-model#the-five-levels)
- Level 3, Tested: Systems are evaluated against defined tests (capability, safety and adversarial evals) and the results are recorded as evidence. Failures are visible, but a failing eval does not yet stop anything. [Chapter 07](https://aigovernanceengineer.com/bok/maturity-model#the-five-levels)
- Level 4, Enforced: The tests bite: policy-as-code and eval gates run in CI/CD and at admission, and a failing control blocks the merge or the deploy. Identity precedes autonomy, and the gating suites’ own coverage and adversarial quality are assessed, not just their existence. [Chapter 07](https://aigovernanceengineer.com/bok/maturity-model#the-five-levels)
- Level 5, Continuous: Assurance is produced continuously from the runtime data path: guardrail decisions, drift and agent behaviour stream into observability, and evidence is emitted as machine-readable artefacts as the pipeline and runtime operate. The audit is a query. [Chapter 07](https://aigovernanceengineer.com/bok/maturity-model#the-five-levels)

## Read more

- [The role in full: chapter 06](https://aigovernanceengineer.com/bok/the-role)
- [The program the role works inside: chapter 12](https://aigovernanceengineer.com/bok/governance-program)
- [The field the role serves: what is AI governance?](https://aigovernanceengineer.com/ai-governance)
- [The Thesis](https://aigovernanceengineer.com/thesis)