Controlled workbook-to-board-pack pipeline

Finance reporting automation Capability project

Built and tested a workbook-to-HTML reporting process, then developed the contracts and controls for reuse across different pack types.

The problem

Manual pack production did not guarantee source reconciliation, fresh evidence or clear interpretation by non-technical readers.
My role

Built a historical pack pipeline around checked workbook data and a reusable data contract.
Where it stands

A historical pack passed 71 pipeline tests and 15 business checks and produced an offline report of roughly fifty-eight pages. The wider reusable factory remains under development; these recorded checks do not certify its current state.

Results and evidence

71 / 71

Historical pipeline checks

Recorded pack build; not a current whole-system result

15 / 15

Historical business checks

~58-page offline PDF

Historical output

Makes the route from checked data to a readable decision pack explicit and reusable.

01
Applied control-total, source-freshness, offline-rendering and reader-comprehension checks to that pack.
02
Separated the checked data contract, view specification and renderer. Engagement analysis and acquisition judgement remain outside the factory.

Tools and working context

A stack-agnostic data contract feeding an offline, confidential-by-design pack generator.

Tools: Python · Pure-Python HTML/SVG renderer · YAML/JSON contracts · pytest · Headless-Chrome PDF
Business area: Board reporting · Decision support
Skill type: Automation · Finance

Failed closed when source evidence was stale or unverified.
Required extracted figures to equal the source statement, preventing internally consistent but incorrect packs.

A real-reader comprehension gate catches decision failures that arithmetic checks cannot.