I lead engineering teams that build AI platforms. My work sits where agent runtimes meet governance: capability control, evaluation, and audit evidence that a regulator can actually read. Previously led engineering for a global logistics platform running across 130 countries. Patent application pending on agentic AI for regulated workflows.
ISO/IEC 42001 Lead Auditor · ISO/IEC 27001 Lead Auditor · AWS Solutions Architect · Azure Solutions Architect Expert · TOGAF
Personal-time work, built in the open. These are reference implementations and working tools, not production software.
Governed agents
| Repo | What it does |
|---|---|
| agentic-os | Ontology-driven agentic operating system as a reference implementation: governed capabilities, identity and budgets, policy in the runtime. Runs offline. |
| agentic-harness | Agent harness with autonomy budgets, blast-radius controls, and audit trails a regulator can read. |
| agent-governance-kit | Audits agent run traces against their governance claims. Findings map to EU AI Act Articles 12 and 14 and ISO/IEC 42001 Annex A. |
| compliance-as-code | ISO/IEC 42001 Annex A and EU AI Act Articles 12 to 15 expressed as code, with an SPDX 3.0 AIBOM emitter. |
LLM and API infrastructure
| Repo | What it does |
|---|---|
| llm-gateway | Multi-provider gateway with an OpenAI-compatible API, routing, fallback chains, budgets and usage accounting. |
| apiloom-ce | API gateway in Go: routing, auth, rate limiting, and a unified LLM proxy. |
Operations tooling
| Repo | What it does |
|---|---|
| sre-copilot | Alert triage and runbook execution behind approval gates, with a full audit trail. |
| confighub | Schema-validated configuration service: environment overlays, secrets indirection, git-backed history. |
Practice and learning
| Repo | What it does |
|---|---|
| Claude-Cyber-Skills | 754 structured cybersecurity skills for AI agents, mapped to MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND and NIST AI RMF. |
| agentic-engineer-path | Onboarding path for engineers joining an agentic AI team: five stages, exercises, and a first 90 days plan. |
| aidlc | An operating manual for AI-supported software development. Tool agnostic, measured per squad. |
| engineering-standards | Engineering standards and checklists, phase by phase, for teams working under regulatory pressure. |
| awesome-ai-governance | Curated list for building and operating AI systems under EU AI Act, ISO/IEC 42001, NIST AI RMF and the UK AI Cyber Code. |
| awesome-rust-ai-governance | Rust crates and projects for AI governance, observability and runtime safety. |
More at mindfulcto-labs.
- Governed agent runtimes and the evidence they leave behind. Capability binding, autonomy budgets, and run traces that answer an audit question without a human writing the answer.
- OpenTelemetry GenAI semantic conventions for agent traces. Getting agent runs into a shape that standard observability tools can read, so governance evidence and operational telemetry are the same data.
- Writing. Turning what works and what fails into notes other engineering leaders can use.
themindfulcto.com covers AI governance, agent safety, and the operating model for engineering teams working with AI.
Open source contributions · LinkedIn · ORCID · Google Scholar
London, UK