Four projects, four sectors, one way of working
A sample of the 100+ cross-sector engagements behind the numbers on this site — chosen because each one is quantifiable, and because I have permission to share it. Figures below are as accurate as I can state them from memory of each engagement; company names appear only where I'm able to name them.
AI-powered prescription fraud detection
A German health insurer (named internally; kept anonymous here for client confidentiality)
The problem: the insurer had to check around 15 million prescriptions a month for fraud — forged stamps, missing or copied signatures, prescriptions issued outside proper authorisation. With a team of roughly 200 staff doing this manually, they could review only around 200,000 prescriptions a month — about 1.3% of the total. Everything else went unchecked.
What I built: an image-processing AI system, designed and delivered as a pilot, that reads each prescription image and checks it against the specific markers of a genuine one — the presence and authenticity of the stamp and signature, and the encryption/security features a real prescription carries. Anything that fails those checks gets flagged for human review rather than approved by default.
The result: the pilot processed the full monthly volume of roughly 15 million prescriptions in about five days, at over 95% validated accuracy — moving the insurer from checking 1.3% of prescriptions to checking effectively all of them, at a fraction of the manual cost.
Robotic-arm formulation testing at the Unilever Materials Innovation Factory, solved for £500
Unilever — Materials Innovation Factory
The problem: Unilever used a robotic arm to test different toothpaste formulations by pressing an indentation tool into paste samples in a grid of dishes, each combination generating a depth reading and a photo of the indentation. The depth told them how good a given formulation was — but the team had lost the ability to reliably tell which generated photo belonged to which tool and which formulation. Without that link, the depth data was unusable. A manufacturing partner quoted around £150,000 to rebuild the process.
What I built: a lower-cost automation prototype that re-established a reliable link between each generated photo, the tool that produced it, and the formulation being tested — restoring the ability to read indentation depth against a known formulation with confidence.
The result: the issue was resolved for £500 — a small fraction of the ~£150,000 quoted elsewhere — and the formulation-testing workflow was usable again.
IoT-optimised haematology laboratory
Royal Liverpool University Hospital, NHS Haematology Laboratory
The problem: the lab's blood analyzer machines process around 2,000 specimens a day, every day. Running that equipment at full capacity around the clock covers demand but wastes energy; running it too conservatively risks missing daily clinical demand. Nobody had the operational data to know where the real optimum sat.
What I built: a bespoke IoT platform that monitored the laboratory for five months, collecting over 50 million real-time operational records. I then applied the Deep Business Analytics (DBA) framework I developed for my PhD — LSTM neural networks combined with a Balanced Scorecard model — to identify the equipment and scheduling arrangement that minimises energy consumption while still meeting full daily specimen demand.
The result: a data-backed operating model that balances energy efficiency against throughput, validated on the full 50M+ record dataset — the same DBA framework that now underpins the predictive-governance work I do with clients. More on the DBA framework →
Coordinating a $200M World Bank-funded programme
World Bank-funded programme
The problem: a $200M World Bank-funded programme involving multiple international and national companies and organisations, each with its own priorities, reporting lines, and delivery pace — needing to move as one coordinated programme rather than a set of disconnected work-streams.
What I did: led coordination and alignment across the full set of stakeholder organisations — international and national — keeping delivery, reporting, and priorities synchronised across the programme.
The result: a multi-stakeholder programme kept aligned and moving as one, at the scale end of the £12k–£200M project range that runs through my career — the same programme-delivery discipline (PMP®/PMI-ACP®/PMI-PBA®-grounded) I bring to engagements of any size today.
Let's talk about what a pilot would look like for you.
These four are a sample, not the full list — if your challenge rhymes with fraud detection, formulation testing, IoT optimisation, or multi-stakeholder programme delivery, a 20-minute call is the fastest way to find out if I can help.
Backed by 25+ years of business leadership — PhD (AI-Driven Business Analytics) · MBA · ISO/IEC 27001 Lead Auditor.