From reporting the past,
to forecasting the future.
I help organisations running complex capital programmes turn delivery data into predictive, AI-driven decisions: governance, foresight and operational control. Proven in telecoms; built for any asset-intensive sector.
Most delivery looks backwards. I build the systems that look forward.
Status decks and variance tell you what already happened. The value is in foresight: predictive models, scenario simulation and decision architectures that tell you what's coming, what it costs, and exactly what to do about it.
AI-enabled delivery intelligence
Forecasting, anomaly detection and ML-driven automation embedded directly into delivery, cutting manual analysis, accelerating decision cycles and improving data reliability at scale.
Decision & insight architecture
Governed data pipelines, a single source of truth and executive-grade dashboards, turning fragmented operational and financial data into decisions shareholders and leaders can act on.
Delivery governance & operating model
PRINCE2- and MSP-grade governance for high-value, asset-intensive programmes: operating models, KPI frameworks and assurance that hold up under shareholder and regulatory scrutiny.
Scenario & investment modelling
Financial and operational models that test decisions before they are made: what-if scenario simulation, capacity and cost modelling, and investment cases that stand up to board and shareholder scrutiny.
From raw delivery data to confident foresight.
Diagnose
Map the delivery landscape, data sources and decision points. Find where reporting ends and risk begins.
Architect
Design governed pipelines and a single source of truth: standards, dependencies and transformation logic built to forecast.
Automate
Deploy predictive models and AI/ML automation for forecasting, anomaly detection and reporting, at programme scale.
Embed
Build data literacy and insight-led ways of working so foresight sticks across teams, suppliers and shareholders.
Twelve years, measured.
Outcomes from more than a decade steering complex, regulated programmes in telecoms, for and alongside Ericsson, MBNL, BT/EE and Three. The methods transfer to any asset-intensive delivery environment.
Running the programme. Building the intelligence that runs it.
Most programme managers hand a brief to a data team and wait. Most AI people have never owned a delivery date or sat in front of a shareholder. I do both, on the same programme, at the same time. Current work inside a regulated, shareholder-governed network operator, described in general terms.
A live command centre for a major capital programme
Leadership was steering a multi-hundred-site programme off monthly decks. I designed and built a browser-based decision platform that gives them live status, milestone monitoring and risk tracking in one place, then made it the programme's weekly governance view.
Portfolio risk forecasting that plans to the downside
Survival modelling with Monte Carlo simulation across 1,000+ at-risk assets. The output is a distribution, not a number, so the programme plans to P90, the bad-but-plausible case. It now underpins funding and prioritisation conversations with shareholders.
On-premises AI agents for regulated data
Local language-model agents (Ollama, n8n) that read and classified over 20,000 public planning records with full data residency and zero per-call cost. Chosen deliberately: it is the governance case for running AI locally inside a shareholder-firewalled business.
AI capability strategy and data classification
An options paper for the leadership team, built on a three-tier data classification (cloud-eligible, shareholder-attributable, security-critical) that reconciles cloud ML with on-premises AI under competition-law and transition-agreement obligations. It scopes the MLOps MVP that follows.
Executive reporting that rebuilds itself
The reporting estate for heads of function and directors, weekly delivery packs, monthly risk reporting and regional benchmarking, turned into one-click pipelines that rebuild from source data. Nothing drifts, nobody spends Friday in a spreadsheet, and the team gets its time back for delivery.
Orchestrated agents that run a delivery process
A multi-agent architecture, an orchestrator with specialist agents beneath it, designed to run a delivery process end to end: correspondence, milestone monitoring, risk resolution. The design test is simple: can a delivery manager see whether an action on a site moved a metric?
Computer vision for field assets
Image recognition that reads field photographs, identifies the physical assets in them and grades their maintenance condition, so inspection scales without sending more people to site. Positioned as the first production workload for the MLOps platform rather than another pilot.
Milestone and energy forecasting the team runs itself
A milestone-attainment forecast that accounts for attrition through to financial year-end, and a site energy consumption model reusable across the business's AI use cases. Both handed over as tools the delivery team runs without me.
A scientist's instinct, applied to delivery.
Insights and Improvement Manager at MBNL, where I run the insight and decision architecture for one of its largest capital programmes and lead its data and AI capability. Before that, six years as a Senior Customer Project Manager at Ericsson, owning high-value, contract-critical programmes for UK operators. A trusted partner to executive and shareholder stakeholders, known for predictive models, executive dashboards and insight architectures that inform decisions from the operational floor to portfolio level.
A physicist by training and an electrical engineer by postgraduate study, I combine enterprise governance with data-driven, AI/ML-enabled decision systems, so leaders stop reacting to the past and start steering toward what's next. My Distinction MSc at UCL centred on neural networks: the very techniques now reshaping how delivery is forecast and governed.
Ready to forecast,
not just report?
Book a discovery call to scope where AI-enabled delivery could give your capital programmes foresight and control.