← Back to the overview

Work, selected · Professional experience

The thinking, in practice.

Pick a case. See my contribution, then try one of the decisions behind it.

01 / Business-led application design

A familiar way to work.
A better way forward.

I designed a shared application around an existing spreadsheet process, keeping familiar editing while connecting records and making calculations explicit.

My contribution · Discovery · Solution design · Delivery & testing

What happenedThe steering group unanimously endorsed the solution, and the project progressed.

Try a design decision

What needs to feel familiar?

A fictional planner: all names, records, calculations and access roles are invented.

“Do we have to learn everything again?”

Change an item’s progress, then switch views. It is the same record in both.

A fictional work planner
Work itemProgressPriority

Two views. One set of records.

The need, the choice and the check
The business need
Let people keep a recognisable way of working.
The design choice
Keep the editable grid, with connected records underneath.
The acceptance check
Change a progress value, then switch views. The same item should keep the same value.

My contribution & delivery details

Understand the work

I studied how the team used its existing spreadsheet and listened to the pain points.

Keep what is familiar

I preserved recognisable editing while connecting records, automating repeated calculations and considering access needs.

Explain the change

I presented what the application would change in day-to-day work and supported delivery and testing. The steering group endorsed it; this is not a claim of production go-live.

Choose another case ↑Next: Follow the dependencies

02 / Data migration assurance

A table is rarely
just a table.

During a data platform migration, I built metadata queries, dependency diagrams and tests around reporting requirements.

My contribution · Mapping · Reconciliation · Investigation · UAT & cutover support

What changedThe diagrams helped the delivery team identify dependencies and plan validation.

Try a dependency

Choose an object. Follow the change.

Illustrative map and tests, independently created for this site.

Inputs
Shared transformation
Outputs
Selected objectFeeds this objectDepends on this object

One shared table, several questions.

Both reports depend on a shared transformation. Testing its inputs and its outputs helps connect technical checks to the questions people ask.

My contribution & delivery details

What I built

SQL metadata analysis, dependency visualisations and matrix-style tests, using report requirements to guide comparisons across SQL Server and Databricks.

What it informed

I shared the dependency analysis with the delivery team to inform validation and investigation.

Where I contributed

Mapping, reconciliation, discrepancy and defect investigation, UAT support and production-cutover activities within the wider programme.

Choose another case ↑Next: Handle the awkward cases

03 / Business-process automation

The happy path is
only the beginning.

I sat with the business user, understood their recruitment process, then built an Outlook, Power Automate and SharePoint workflow around it.

My contribution · Discovery · Workflow design · Testing · User guidance

In useThe recruitment workflow. A related adaptation later reached a tested endpoint.

Try an exception

Give the workflow something awkward.

Guided illustrations of the design questions; no workplace system is connected.

  1. 01 · Check the input
  2. 02 · Preserve the documents
  3. 03 · Confirm completion

Two attachments. The same filename.

Two different applications each include a file called CV.pdf.

What could go wrong?

Saving both as CV.pdf in one location could overwrite a document or mix up applications.

The design decision

Use separate application destinations so both documents survive.

Make it a test

Check that both CVs remain available and each belongs to the correct application.

My contribution & delivery details

In use

The recruitment workflow is in use. Discovery, design and deliberate testing helped connect the automation to the team’s familiar process.

Adapted and tested

I later adapted the approach for a related workforce process, reaching a successful tested endpoint.

Made understandable

I translated the flow into plain-language visual guidance and walked users through it. Delivery included helping people understand what would happen and why.

Choose another case ↑Next: Make the next step clear

04 / Visual documentation

Useful instructions.
A local tool.

IT procedures needed clear screenshots without sending account details or access settings to an external editor. I built a reusable PowerPoint annotation toolkit.

My contribution · Requirements awareness · Toolkit design · Wiki guidance

The deliverableAn editable toolkit, with reader and creator manuals for the team Wiki.

Take the toolkit

Clearer instructions. Yours to keep.

Paste a screenshot. Add markers, arrows, notes and covers. Export a clear instruction.

The PowerPoint toolkit is the original approach. The browser editor uses a fictional sample.

Download PowerPoint toolkit Try the browser editor

Use the .pptx in desktop PowerPoint. An offline browser edition is also available in the studio.

Example workspace
1
Account detailsReview request

Show the action.
Keep the private details covered.

My contribution & delivery details

The requirement

Make visual instructions easy to create while keeping sensitive screenshots within the local editing process.

The choice

Use a familiar tool colleagues already had. Reusable markers, arrows, highlights and notes gave authors a consistent starting point.

The handover

Keep the working slide editable, then export a single annotated PNG for the Wiki. Reader and creator manuals explained how to use and maintain the documentation.

Choose another case ↑Next: Explore application design

These accounts describe my contribution. Interactive examples are independently created; internal names, records, diagrams and code stay private.

Build with a real public dataset