EdTech Platform Deployment Across Southeast Asia
Rolling out integrated learning management and assessment systems to schools across four countries — with six-month delivery windows per territory.
Context
The organisation had a learning management platform serving a single market. The technology worked, but it had been built for one country's curriculum, one language, and one regulatory framework. The mandate was to expand across four countries in Southeast Asia within two years — each with different curriculum standards, different languages, different data residency requirements, and different expectations for what a "school platform" should do.
This wasn't a product rollout. It was four concurrent localisation projects, a platform architecture overhaul, and a sales motion — running simultaneously, with a team that didn't exist yet.
Approach
The first decision was sequencing. The commercial pressure was to launch in all four markets simultaneously. The technical reality was that the platform couldn't support it — the data model assumed a single curriculum structure, and the deployment pipeline was manual.
The hardest conversation in any scale-up project is the one where you tell the commercial team that going slower now means going faster later. You have to be right, and you have to be specific about why.
I proposed a phased approach: lead market first (closest to existing architecture), then two parallel markets, then the fourth. Each phase would force architectural decisions that the next phase would inherit. The lead market was the proving ground — not just for the product, but for the deployment playbook.
Building the team happened in parallel. I hired across engineering, QA, customer success, and implementation — growing from a core of five to twenty-five people over eighteen months. The hiring strategy was deliberate: senior engineers who could make architectural decisions independently, mid-level implementers who could execute the deployment playbook, and customer success people who understood education (not just SaaS).
The platform architecture had to change fundamentally. We moved from a single-tenant monolith to a multi-tenant system with curriculum-aware data models, per-territory configuration, and localised assessment engines. This was the highest-risk workstream — any architectural misstep would multiply across every subsequent deployment.
Outcome
All four markets launched within the two-year window. The platform scaled to over 150,000 monthly active users — teachers, students, and administrators — across 450+ schools.
The phased approach paid off in a specific, measurable way: the lead market deployment took fourteen weeks. The second and third markets, running in parallel, took nine weeks each. The fourth took six. Each deployment refined the playbook and exposed configuration requirements that became template defaults for the next territory.
The team grew from five to twenty-five with zero involuntary attrition during the scale-up period. Two of the senior engineers I hired went on to lead their own platform teams within the organisation.
Reflection
Three things I'd carry forward from this project:
Sequence for learning, not for speed. The phased approach was slower on paper but faster in practice because each phase eliminated uncertainty for the next. The temptation to parallelise everything is strong when there's commercial pressure, but parallel execution without shared learning just means you make the same mistakes in four countries instead of one.
Hire for the role after next. The senior engineers I brought on weren't just solving today's problems — they were making architectural decisions that would constrain or enable the platform for years. Hiring for current-state needs in a scale-up project means you're always behind.
The deployment playbook — the boring operational document that nobody wanted to write — turned out to be the single highest-leverage artefact of the entire programme. Every week of process documentation saved three weeks of deployment time downstream.
Invest in the playbook, not the heroics. The first deployment relied on senior engineers doing things manually, adapting on the fly, solving problems through skill and effort. That's fine for one deployment. It doesn't scale to four. The work of turning tacit knowledge into repeatable process — writing the playbook — felt slow and bureaucratic in the moment. It was the most important work we did.