About this site
What this is, who maintains it, and how to contribute.
AI Governance Engineer is the home of AI Governance Engineering: The Thesis & Body of Knowledge, a founding text and working reference for turning AI governance obligations into policy-as-code, eval gates, agent registries and machine-readable evidence. It treats governance as a system to be built, run and measured, not a document to be signed. If you are new to the field, start with what AI governance is.
Who maintains it
Written and maintained by Jorge García Aibar, AI Governance & Privacy Engineer, sole author of the Body of Knowledge, website, datasets and project materials. The Thesis is the sole exception and is co-authored by Jorge García Aibar and Aurélie Pols. This is a public draft and invites practitioners who build governance for AI systems to propose improvements.
About the author
- Name
- Jorge García Aibar
- Role
- AI Governance & Privacy Engineer
- Experience
- Two and a half years designing and operating an AI governance framework inside a large telco, sitting between Legal, Security and Engineering, and covering the governance, security, compliance, business and model-performance dimensions of AI risk.
- Profiles
- LinkedIn · GitHub
Research interests
- Independent assessment methodology for AI systems
- Evaluation-environment controls
- Agent runtime safeguards
- Machine-readable assurance evidence
- Open control profiles and reference implementations
Current open questions
The questions I am working on and have not settled. Each links to where the work on it lives; a question leaves this list when a reviewed note or control answers it.
- What evidence is sufficient to demonstrate that an agent remained within an authorization boundary?
- How should evaluation-harness configuration be attested?
- Which runtime safeguards should fail closed?
- How should control regressions be detected over time?
- How should machine-readable assurance evidence preserve provenance?
Open work
What is in progress, with its real state. Nothing here is marked reviewed until a named reviewer is credited; until then each item is open for technical review.
- Evaluation Environment Control Profile
Nine draft control specifications for the environment a model or agent is evaluated in; open for technical review.
- Agent Runtime Control Profile
Forty-two reference controls derived from the agent controls of chapter 23; every control is a draft.
- The evaluation environment is part of the system
A research note on why an evaluation that ignores the harness, tools, credentials and network measures only part of the system.
- Frontier labs and evaluators route
Where the controls, patterns and evidence formats on this site apply to evaluating and deploying frontier systems.
- Incident notes on three cases
System boundary, control assumptions and the controls that would have caught each failure, for three incident cases.
- Four planned research notes
Themes listed on the research index and not yet written.
Let’s work together
I’m Jorge García Aibar. I am particularly interested in collaborations around independent assessment methodology, evaluation-environment controls, agent runtime safeguards, machine-readable evidence, open control profiles and reference implementations. If you are building or assessing any of these, compare notes with me.
Email me at jorgegarciaaibar@gmail.com or contact me on LinkedIn. For public questions or proposals about the Body of Knowledge, start a GitHub discussion.
How to contribute
Everything here lives in a public repository. Propose a change, add a pattern, sharpen an argument or fix a fact by opening a pull request. Every factual claim needs a sourced, verified citation, following the style guide. Accepted substantive contributions are credited by name; they do not change the stated authorship unless expressly agreed.
- Methodology: how sources are chosen and tagged, how dates are kept current, and how corrections, reviews and releases work.
- Contributors and signatories: who has written, revised or signed.
- Changelog: what changed in each version.
- Source repository on GitHub: open an issue or a pull request.
Download
The Body of Knowledge as released in v0.5.0 (the Thesis, every chapter of that release, contributors and changelog) is available as a single PDF. Download the PDF (v0.5.0, 4.6 MB).
Stay updated
New and updated chapters, and each released version, are announced in the RSS feed.
Cite
Each release is archived on Zenodo. This version (v0.5.0) has the DOI 10.5281/zenodo.22956197; the concept DOI 10.5281/zenodo.22857084 always resolves to the latest one. Machine-readable metadata lives in CITATION.cff; every chapter offers a ready-made citation and BibTeX entry.
Licence
Except where otherwise noted, this work is licensed under CC BY 4.0. You may share and adapt it with appropriate credit, a link to the licence and an indication of changes. Attribute the Body of Knowledge, website, datasets and project materials to Jorge García Aibar. Attribute the Thesis to Jorge García Aibar and Aurélie Pols.
Contact
Email: jorgegarciaaibar@gmail.com. Discuss the discipline on LinkedIn, or on GitHub.