Finance systems, Anaplan architecture and applied AI. CIMA-qualified accountant and Master Anaplanner, with fifteen years in FP&A and planning systems. I build tools that help finance teams understand their models, review changes and trace an AI-generated answer back to its evidence.
I'm interested in roles combining finance domain knowledge, solution architecture and hands-on delivery. The projects below show the work: working examples, implementation choices, tests and the limits of what has been validated.
My work spans finance operations, planning architecture and implementation across biotech, pharma and other reporting-heavy organisations.
- Close and forecasting: at a listed clinical-stage biotech, I built the Anaplan planning platform and automated its data loads. The mechanical close fell from 12 working days to under 8 hours, and reforecasting from a week to a day. Here, mechanical close means ledger close through to consolidated actuals ready for review. My delivery case study explains the scope, integrations and handover.
- Professional background: CIMA-qualified Management Accountant, Master Anaplanner and MSc in Information Technology from Keele University. Background and experience.
- Other perspectives: Anaplan featured my career and approach in Meet Solutions Architect Karim Lameer. LinkedIn recommendations from colleagues describe my Anaplan delivery, financial understanding and ability to become productive quickly in a team.
| Project | Problem it addresses | Evidence to inspect |
|---|---|---|
| Anaplan Estate | What should we investigate in an inherited estate, and what could a change affect? | Try the report, case study, validation limits. |
| Anaplan Grammar | How do we analyse formula structure and dependencies reliably? | Parser and graph, public regression tests, engineering walkthrough. |
| The Audited AI Close | How can an assistant coordinate a month-end close with calculation checks and human review? | Case study, calculation scripts, finance-team runbook. Fictional data. |
| The Board Pack Test | Can an AI system answer questions across realistic finance documents and supply the right sources? | 34 documents, 25 questions, saved answers and versioned grading checks. Results distinguish automatic checks from human judgment. |
| Grounded field notes | What does it take to operate a finance document assistant? | Case study, architecture decisions, incidents, my contribution and upstream work. Documentation only; application code is private. |
- Start with the finance process, its users and the decision the output must support.
- Make calculations and source references inspectable; keep human judgment explicit.
- Test against known answers and failure cases, and record what remains unproven.
- Explain the architecture, rejected alternatives and operational consequences.
The Anaplan parser was developed against 13,214 unique private formulas. The public tests use fictional formulas; the private corpus result is not independently reproducible from this GitHub. Estate recommendations still need validation on unseen estates. The AI close is a fictional reference workflow, and the Board Pack Test covers one company. Each repository gives the evidence and scope behind its claims.
Anaplan Diff · Impact Explorer · API Starter · Anaplan Clock
LinkedIn — experience and contact · CodelessOps — projects and writing