From AI to infrastructure. Systems I was trusted to change.
AI-assisted operations, networks, 15,000-rack distributed data-centre estates, regulated resilience, cloud and enterprise operating models – at programme scale.
The technology changes. The mandate does not: connect architecture, automation, operations, security, risk and economics until the system becomes measurable, governable and capable of acting with less human friction.
AI-assisted / closed-loop operations programme context
data-centre racks under transformation scope
European telecom data centres
employees affected by a major carve-out transformation
Different technologies. The same executive problem.
AI may be the headline today, but production AI still lands on infrastructure, networks, data, security, operating models and economics. My work has repeatedly sat at those joins – from AI-assisted autonomous operations to telecom backbones, data-centre telemetry and regulated resilience.
AI-Assisted Autonomous Operations & Closed-Loop Observability with self healing
A global luxury enterprise needed to move beyond conventional monitoring: complex Kubernetes-based platforms were generating enough operational telemetry to see problems, but not yet to close the loop between detection, diagnosis and corrective action.
Design a monitoring and observability model that could support increasingly autonomous operations – combining platform telemetry, automation, self-healing patterns and AI-assisted analysis rather than simply creating more dashboards and alerts.
Built closed-loop monitoring and observability patterns around Kubernetes, automated remediation and self-healing workflows, with AI incorporated into the operational chain to improve detection, analysis and response. The emphasis was on making observability actionable: telemetry had to trigger a governed operational response, not another ticket queue.
The operating model shifted from reactive monitoring toward closed-loop, AI-assisted operational control – creating the foundation for more autonomous platform operations.
programme context
operations & analysis
platform foundation
observability + self-healing
Global Data Centre Intelligence & Infrastructure Economics
A global data-centre estate of roughly 15,000 racks was being managed through fragmented infrastructure information. Hardware lifecycle, rack utilisation, thermal behaviour, electrical consumption, software density and workload placement were visible in isolation, but not as one economic system.
Transform data-centre management, governance, monitoring and observability so that physical infrastructure telemetry could be connected all the way through workload consumption to service economics and COGS.
Built a custom monitoring, observability and reporting capability using open-source tooling. The model connected hardware lifecycle, rack temperature, power consumption, electrical phase balance, rack density, server/software density and workload allocation. Under-utilised racks could be consolidated, over-dense racks redistributed, cooling parameters corrected and abnormal power consumption used to expose equipment or infrastructure faults.
The estate moved from infrastructure monitoring toward a decision system for density, thermal efficiency, power, lifecycle and workload economics – the same telemetry foundation modern AI-driven operations depend on.
programme context
racks in scope
efficiency gain in targeted areas over 12 months
custom observability stack
European Telecom Backbone Virtualisation & Dark-DC Automation
Critical telecommunications network functions were historically coupled to dedicated physical network equipment across a large European data-centre footprint, creating hardware dependency, fragmented capacity and high lifecycle complexity.
Design the infrastructure, governance and automation model for virtualised network functions forming part of the European telecommunications backbone across 125 data centres.
Decoupled network functions from dedicated appliances, introduced virtualised infrastructure, automated provisioning and workload distribution, and designed an operating model capable of supporting dark data centres with minimal routine human presence.
The transformation materially reduced the physical network estate while increasing automation and operational consistency across the European footprint.
programme context
European data centres
reduction in network equipment
automation-led operating model
Infrastructure Consolidation & Service Repatriation
A large mobility-services environment had become distributed across 17 external providers, including cloud services, creating cost, governance and operational complexity across a user base of roughly 25 million.
Consolidate the estate, redesign governance and operations, and determine which services should remain external, move to cloud, or be deliberately brought back under direct control.
Designed a four-data-centre target estate, consolidated services from 17 providers, rationalised service ownership and operating processes, and applied an economics-first rather than ideology-first approach to placement decisions.
The result was a substantially simpler estate with lower cost, stronger control and a more coherent operating model.
users
provider / DC consolidation pattern
cost & complexity reduction
target operating estate
Enterprise Carve-out & Digital Workplace Transformation
A major corporate carve-out required workplace services, governance and IT operating structures to be separated and rebuilt without losing continuity for approximately 60,000 employees.
Move workplace systems into a new cloud-based model while establishing the governance, operating structures and accountability needed for the separated organisation.
Led transformation across workplace, cloud, governance and operating-model domains, separating inherited structures, simplifying service ownership and redesigning the environment around the needs of the carved-out organisation.
The transformation reduced both cost and structural complexity while enabling the separated organisation to operate on its own technology foundations.
programme context
employees
cost & complexity reduction
new governance & IT structures
Greenfield 5G Infrastructure Automation
A new mobile network required infrastructure to be built at scale without inheriting the manual provisioning, configuration drift and operational overhead associated with traditional telecom estates.
Create an infrastructure-as-code and automation model covering deployment, governance and operations for a greenfield 5G environment.
Applied IaC principles, automated provisioning, repeatable configuration, governance controls and an automation-led operating model designed for a software-defined network environment.
The programme established automation and repeatability as first-class design principles rather than retrofitting them after the network was built.
programme context
greenfield network
infrastructure delivery model
operations & governance
Operational Resilience & Critical Infrastructure Transformation
A highly regulated financial-market environment required a secondary operating capability that could remain secure, controlled and recoverable during a severe primary-environment disruption.
Provide cross-functional transformation leadership across a control surface normally split between CIO, CISO, infrastructure, architecture, operational risk and programme governance.
The mandate spans resilience architecture, security controls, IAM/PAM, privileged access, immutable backup and restore, supplier governance, operational readiness, service levels, dependency management, executive decision preparation and regulatory alignment.
The mandate brings architecture, security, operational risk, suppliers and recovery readiness into one governed system rather than treating them as separate workstreams.
recovery objective
cross-functional scope
operational resilience
market infrastructure
Different problems. The same operating discipline.
Not every engagement needs a full case study. These illustrate the breadth behind the flagship transformations.
Enterprise Architecture & Service Economics
~€230M programme context. TOGAF-based architecture, showback / chargeback and governance linking technology services to production, supply chain, delivery and business capability.
Agentic AI, RAG & AI Infrastructure Patterns
Hands-on work underneath the AI layer: agentic workflows, MCP, RAG, vector databases, orchestration and policy/governance patterns – with particular focus on how AI platforms connect to enterprise infrastructure and operations.
SaaS Economics & Reliability Transformation
Shared services across Platform, DevOps, SRE and Observability with direct P&L impact: €10M+ annualised savings, −26% COGS/tenant, +6 gross-margin points and −78% MTTR.
R&D Portfolio & Delivery Operating Model
Global programme office and delivery governance across 72 FTE / 24 countries, improving milestone on-time performance from 67% to 92% and delivering €3.6M annual run-rate savings.
Data Centre Build & Infrastructure Engineering
Selected experience also includes co-building major enterprise data-centre environments in Europe, giving the transformation work a grounding in the physical engineering layer as well as architecture and operations.
From AI agents to power, racks and risk.
The technologies span AI platforms, autonomous operations, 5G, cloud, data-centre telemetry, workplace systems and regulated resilience. The underlying management problem is remarkably consistent: the estate becomes too complex to govern through organisational silos.
My role is to connect platforms, infrastructure, workloads, suppliers, security, risk, operations and economics into a system that executives can actually control – and increasingly, one that can automate more of its own operation safely.
If your situation resembles one of these, start with the problem - not a sales deck.
Bring the transformation, decision or operating problem. We will establish whether my experience is relevant and what kind of mandate would make sense.
Due to the confidential nature of board- and C-level mandates, client identities, sensitive architectures and commercially material details are intentionally anonymised. Scale and outcomes are shown only in aggregated or approximate form. References can be discussed directly where permissions allow.