Work  /  Think Tank
Case study · Think Tank Training Centre

Scaling a top-ranked academy without breaking what worked.

Think Tank was already globally ranked. My team and I rebuilt the learning operating system underneath it to scale without losing the standard.

95%+
On-time completion
sustained
3×
Scale across learners,
teams & content
Top 10
Global, every area of study
· The Rookies
75%+
AI scaffolding
time saved
Summary

I joined at a turning point.

The reputation was strong. The system underneath wasn't ready to scale: metrics too soft to drive decisions, the same quality issues resurfacing, content lagging demand.

My team and I rebuilt the foundation, scaled the engine, and held the standard across learners in 50+ countries.

The brief
01

Strengthen the operating system before scaling.

02

Modernize the content pipeline.

03

Hold quality through growth, across an international cohort and under provincial regulatory oversight.

Built on

How it was done

Three moves, in order.

Two patterns kept surfacing in the quality data: content drift and support saturation, neither measured well enough to fix at scale. Measurement came first.

  • StandardsQuality governance and weekly review across all 77 courses, aligned to PTIB / PTIRU.
  • MeasurementSeven weekly KPIs for completion, support, and quality.
  • SystemsRebuilt educator onboarding, performance management, and student-of-concern frameworks.

You can't scale what you can't see. I rebuild measurement before anything else.

The bottleneck was production, not talent. We rebuilt the pipeline around AI scaffolding and cut planning, outlining, assessment, and synthesis time by 75%+.

AI drafted skeletons, rubrics, and summaries. SMEs still owned development, scripting, recording, and the final say. Leverage, not authorship.

Content production: traditional vs. AI-augmented
BEFORE Serial pipeline. Every stage manual. SMEs bottlenecked. Needs Manual Design Manual Develop SME-led Produce SME-led Assess Manual Publish & review Manual AFTER AI scaffolds the framing. SMEs still own the craft. Needs AI-assisted Design AI-assisted Develop SME-led Produce SME-led Assess AI-assisted Publish & review AI-assisted AI-assisted scaffolding & synthesis     Human-owned craft
  • CurriculumFull refresh of CG Asset Creation; launched Houdini for Film or Games, a 16-month intensive.
  • Scale25–28 SMEs concurrently, 100+ across tenure; foundations for LMS 2.0 with Engineering and Product.

AI is leverage, not authorship. I rebuild engines so experts spend time where it matters.

Both root patterns cleared. Under the new framework, on-time completion climbed and held at 95%+, sustained across a fully international cohort as the school scaled.

I left when the work was done. The systems are still running.

Content footprint

77 courses, sorted into 5 specializations.

Each square is a course, colored by the stream it belongs to: four under CG Asset Creation, plus the Houdini generalist program. One shared standard underneath them all.

10–14Lessons / course
1,200+Hours total
5Specializations
The real test of an operating system isn't how it runs while you're there. It's that it keeps running after you leave.
From a colleague

"Mel has a rare ability to design and scale operating models that work in practice, balancing strategy, execution, and people with clarity and calm."

Laura Coumbe  ·  Artist Manager & Talent Specialist, Think Tank Training Centre
External validation

The work was measured by the people who hire from it.

Two independent industry sources documented work I led at Think Tank. The Rookies methodology is based entirely on portfolio reviews by working industry professionals, so the rankings measure what the learning systems produced.

Let's talk

Building or rebuilding an L&D function?

That's the work I do. If it's the work you need, get in touch.

Get in touch →
← Back to work
Next case study
Building a function from scratch: CEA →