Adopting AI under SOC 2 Type II, ISO 27001, GDPR and the EU AI Act: the seven things every framework asks about an AI system, how to engineer for them, and what each team in the organization needs.
Badges
Badges come from GitHub once the author claims this feed.
Recent posts
every link opens the author’s post-
Adopting AI inside an organization that has to pass a compliance audit
-
AI security engineering: the same failures, but new ways to attack
Reflections on the last year of AI security incidents: new attack techniques, the same failures they keep exploiting, and why AI security engineering is now a core discipline.
-
Agentic AI engineering: Read-only MCP servers by default, and never run as my own identity
How I fence the tools my coding agents reach for: the Microsoft Learn MCP Server for current documentation, the Azure MCP Server behind a read-only principal scoped to the work, a browser through Playwright MCP, no cloud CLI from the terminal, and the servers I keep out of the…
-
Agentic AI engineering: Multi-agent orchestration across lanes without conflict
How multiple coding agents and their sub-agents share one working tree without conflicts: lanes isolated by files, claiming lanes, getting a set of work items, and messaging between agents to sync on work.
-
Agentic AI engineering: Why instructions aren't enough for coding agents
In this post I explain how agents instructions vs. enforced rules behave, and why it matters.
-
Agentic AI engineering: What the humans do, what the agents do, and where the hard lines are
The split between what my coding agents do and what stays with me, the four ways work comes back to me, and the acts an agent never performs on its own.
-
Agentic AI engineering: I steer, the agents orchestrate
What agentic AI engineering means in practice: one engineer who steers, several coding agents that carry the work, and a board that shows who is doing what and what is waiting on me.