Twenty-five years across defence, pharmaceutical and banking, in roles where one bad change reaches regulators, patients or customers. Now securing and building production agentic AI systems.
I've spent my career inside FDA-validated pharmaceutical systems, security-cleared defence and aerospace programmes, and retail banking: environments where a mistake means a regulatory finding or an outage. In those places you earn trust by being right about risk, year after year.
I'm hands-on by instinct and senior by experience. I've held enterprise-wide administrative authority over critical estates, designed the security controls beneath them, and in recent years built and secured agentic AI systems that run real business processes. That combination of security ground truth and working agentic AI is where I focus now.
Part of a small core team holding Enterprise Admin and Root Forest Admin authority across a global FDA-validated estate, including large-scale Active Directory consolidation.
UK MoD Developed Vetting (DV) cleared. Delivered secure infrastructure for defence and aerospace programmes, including one of the earliest RSA SecurID two-factor deployments in UK defence.
Senior interim infrastructure and security work inside major UK retail banks, in environments with strict change control and audit expectations.
Led cloud migration, Microsoft 365 adoption and integration architecture for a multi-year engagement, and built a working LLM-driven integration between finance systems and the business.
The interesting problems now sit where autonomous AI meets real systems and real risk. I work that seam from both directions: as a hands-on security practitioner, and as a builder who runs agentic systems in production.
Helped unify a fragmented global directory estate into a single, governable forest under strict validation requirements.
Delivered one of the earliest RSA SecurID strong-authentication deployments in the UK defence sector.
Ran Exchange Online administration through migration, including recovery from a retention-policy incident affecting more than 160 mailboxes.
Designed and shipped a working Model Context Protocol server connecting an ERP/finance platform to the business through large language models, then decommissioned it with full documentation.
If you're hiring for the place where security, infrastructure and agentic AI meet, I'd be glad to talk.