Enterprise AI architect and operator, UK-based. I build AI systems and I govern them — most people only do one of those.
SS1/23, Consumer Duty, EU AI Act — most governance content is a PDF checklist. Mine is working code that produces audit evidence. Prove it's governed, don't just declare it.
fs-ai-governance-demo — runtime AI governance for financial services, end to end: OPA/Rego policy enforcement → SHA-256 tamper-evident audit ledger → drift detection, mapped to FCA Consumer Duty, SM&CR, and SYSC. Runnable in minutes; the test suite proves every control does what it claims.
governance-harness-v0 — a runtime governance gateway for regulated AI agents. Four policy gates (data, tool, output, human-approval) sit in front of a mock agent across eight synthetic scenarios. Ungoverned, 14 violations slip through; governed, all 14 are redacted, blocked, or escalated — with a CI release gate that fails the build if any violation goes unmitigated or any eval score drops below threshold. Every decision is a structured, framework-mapped record (EU AI Act, PRA SS1/23, ISO/IEC 42001, NIST AI RMF, FINOS AIGF), producing a real evidence pack: audit trail, control mapping, go/no-go.
iso-42001-ai-governance-pack — production infrastructure for ISO 42001 and EU AI Act compliance, deployable on AWS in under an hour. Terraform (private VPC, KMS, CloudWatch, least-privilege IAM), OPA/Rego policies enforced in CI, NeMo Guardrails for PII/jailbreak/output filtering, and a compliance report generator mapped control-by-control to ISO 42001 Annex A and EU AI Act Articles 9-15.
This is the differentiator: builders who ship agents rarely ship the control plane around them. I do both.
Teaching product leaders, enterprise architects, and business analysts how AI actually changes how organisations run — not slideware, working repos.
copilot-architect-kb— production-ready AI systems and enterprise architecture patterns, Azure and beyondarchitecting-ai-implementations— architectural patterns and best practices for enterprise AI, design through productionkb-implementation-examples— 50+ runnable code examples from the Copilot Architect KBdata-engineering-journey— hands-on course: traditional data → modern engineering → AI-driven systems, built for engineers moving into AIai-tools-platforms-solutions-guide— a guide to AI tools, platforms, and enterprise solutions for architects and consultants
Small, usable tools — not platforms — for non-technical founders adopting AI.
aqua-voice— system-wide voice typing for macOS using Deepgram real-time transcriptionself-learning-browser-automation— browser automation that improves from every interaction using AI memory and reinforcement learning
Book a session → or reach me on LinkedIn / email.