Commercial leadership in complex B2B advisory, supported by practical AI product and cloud-cost work.
I bring 20+ years of commercial and B2B experience, including approximately 12 years in specialist UK R&D tax advisory. At Provident Tax I progressed from the first commercial hire to Director of Commercial, Finance & Operations, built a 12-person commercial function and helped grow annual revenue from approximately £4.8m to £10m.
My work connects customer needs, senior stakeholder relationships, commercial decisions and delivery. This portfolio shows how I apply that approach to product discovery, AI prototypes and technology cost evidence.
Professional background on LinkedIn
| Project | What to look for | Stage |
|---|---|---|
| AWS platform & FinOps | Billing evidence, cost ownership and explicit cost/resilience trade-offs | Portfolio environment and controlled experiments |
| WorkTwin product case study | Stakeholder-led product decisions, governed AI guidance, evaluation and human escalation | Controlled-preview prototype |
| MCP document tools | A focused Python integration connecting local document conversion to AI assistants | Trusted local prototype |
Built an AWS/EKS portfolio environment with Terraform and used committed billing evidence to examine cost, ownership and resilience. Controlled experiments recorded an 83.33% lower EKS control-plane rate and a 50% lower fixed NAT/IPv4 rate. Showback attributed 99.61% of gross usage cost, distinguishing direct tags from rule-based allocation.
These results concern a small development environment. The repository separates implemented features, recorded measurements and modelled monthly savings; it does not claim enterprise-wide savings.
Explored how care staff could find approved guidance and navigate realistic scenarios without duplicating existing care records. Stakeholder feedback changed the product direction towards guidance, source references and appropriate human escalation.
The public case study explains the decisions, prototype architecture, historical evaluation, my AI-assisted contribution and remaining work. WorkTwin remains a supervised prototype; production adoption and commercial outcomes have not been established.
Connected local PDF/DOCX conversion to an assistant through MCP, using Microsoft MarkItDown for the underlying conversion. The repository provides the interface and basic fixture tests, with clear limits on hosted and confidential-document use.
- Consultative B2B selling, business development, strategic partnerships and account growth.
- UK R&D tax advisory, commercial operations, forecasting and governance.
- FinOps Certified Practitioner and AWS Certified Cloud Practitioner.
- Practical AI-assisted development using Python, TypeScript, retrieval workflows and cloud infrastructure.
My professional identity is commercial and advisory. The projects provide additional evidence of product judgement, technical fluency and collaboration with engineering.
Repository history, third-party components, historical test checkpoints and limitations matter. Forks and learning resources are not presented as original flagship builds.