File
Evidence pack
Classification, risks, controls, tests, owners, residual risk.
AI Governance & Compliance
EU AI Act, ISO 42001, NIST AI RMF — mapped onto a real system: roles, evidence, and a file a reviewer can open.
Created by Baljeet Dogra
Course Objectives
- Classify a system and defend the classification.
- Map risks to controls to tests — not to slogans.
- Name owners: who signs, who operates, who is on the hook.
- Build an evidence pack a reviewer can navigate in twenty minutes.
What you will produce
Map
Framework crosswalk
Act / ISO / NIST on one page for your system, gaps named.
4 weeks curriculum
Expand a part for the syllabus. Content stays searchable when closed.
01 The landscape Week 1
Act, ISO 42001, NIST AI RMF, sector overlay. What actually binds.
- Act
- ISO
- NIST
- Sectors
02 Classify and risk Week 2
Use cases, prohibited, high-risk, GPAI. Model risk language.
- Classification
- GPAI
- Model risk
- Proportionality
03 Controls and evidence Week 3
What good evidence looks like. Logging, human oversight, quality.
- Evidence
- Logging
- Oversight
- Quality
04 The pack Week 4
Peer review of packs. Capstone: twenty-minute walkthrough.
- Owners
- Residual risk
- Review
- Capstone
Who this is for
Engineers and architects
You will be asked for a file. You need one that is true.
Risk, legal and consultants
You need enough system literacy to not write unauditable policy.
Prerequisites
- You can describe an AI system in production language
- Comfortable writing for a non-engineering reader
Not a fit if
- A certification dump with no artefact
- People who want to skip the system and “just do policy”
Related: The AI Consultant · AI Security & Privacy
Questions
Is this legal advice?
No. It is engineering evidence against published frameworks. Your counsel still signs.
UK or EU?
The Act as the binding EU text, plus how UK organisations typically map it. You write for the system you have.
Ready to start?
Four weeks to an evidence pack a reviewer can open. Create an account to enrol.
Enrol now