Systems Architect focused on AI governance, failure modes, and recoverable automation
I design systems where AI can propose actions without silently causing irreversible damage.
My work centers on determinism, explicit constraints, auditability, and human-in-the-loop control.
Most of what I build exists to answer one question: "What happens when this system is wrong?"
These artifacts are working documents: playbooks, diagnostic tools, and guardrails used to design and evaluate complex systems.
A governance layer for AI prompting that makes "vibes-based" prompts illegal. This is the core system. The artifacts below are extracted control surfaces or applied examples.
What this shows: How I think about guardrails, accountability, and making AI workflows auditable.
A deterministic file-intelligence system. Enforces explicit proposal vs execution separation, governance checks before any action, journaling, reversibility, and deterministic fallbacks when confidence is low.
What this shows: A practical testbed for AI governance principles under real constraints. Standalone product, referenced here as a systems artifact.
An operational control extracted from The Prompt Governor.
What this shows: How I enforce accountability before execution.
An operational control extracted from The Prompt Governor.
What this shows: How I prevent people from automating chaos and calling it efficiency.
An operational control extracted from The Prompt Governor.
What this shows: How I turn AI "hallucinations" into measurable failure modes.
A production sales-order system redesigned to be observable, testable, and automation-ready — without premature autonomy.
What this shows: How I translate governance theory into real operational systems that humans and future agents can safely execute.
I work at the intersection of systems design, automation, and AI governance.
My background is not in chasing novelty, but in preventing fragile systems—especially those that fail quietly, at scale, and without clear ownership.
I care about:
Most of my work is exploratory by design. Some of it becomes products. All of it is governed.
Email: leo@bbstrats.com
I read everything. Replies may be slow.