Proof

Selected work — honest, not decorated.

Selected contexts. No fake testimonials. Numbers only where I can stand behind them.

Multi-programme delivery under load

Situation

Several QA programmes running in parallel; staffing and deadlines competing every week; quality signal easy to lose when every account feels urgent.

Action

Tightened the operating rhythm: honest capacity, written priorities, clearer risk calls, and protection of focus so testers could do the work that ships.

Result

Delivery that holds when several clocks ring at once — fewer quiet failures (thin roster + unclear priority) before they become Friday emergencies. (Ongoing Directorship context at QATestLab: ~5 programmes, ~140 people.)

A bounded AI loop in the SDLC

Situation

Recurring weekly grind on a large test-case load (400+ cases): hours spent on maintenance and checkable work that did not need senior judgment every time.

Action

Built a four-agent n8n system around that one loop — structured input, checkable output, human spot-check — instead of a broad "AI for testing" programme.

Result

Roughly ten hours/week → about one hour/week on that path; around 30% fewer errors on the automated grind. Useful because the fence was clear. Not a claim that AI fixed delivery.

Selected contexts

Delivery Director, QATestLab (Madrid / EU)
Program Director: delivery organisation scaled ~35 → ~80 (iGaming, PC/mobile games, commercial apps)
Programme delivery management, large fintech client (NDA — unnamed)
Project management, ESL FACEIT (FACEIT platform and tournaments)

If you need a deeper walkthrough of any of these, ask. I won't invent logos or quotes to fill a page.