Chapter 04
The Systems
2021–2024
Enterprise systems for airports and universities, EdTech across four countries, and the first AI-powered features — learning to hold complexity without simplifying it away.
Two companies, one thread
This chapter splits across two Singapore companies, but the thread is the same: complex, multi-stakeholder systems at institutional scale.
At WeesWares, I managed projects as a development studio PM. At CommonTown, I managed products. Both roles demanded systems thinking — the ability to see how components connect, where dependencies hide, and what breaks when you change one thing.
WeesWares: enterprise systems
WeesWares was a development studio serving major institutional clients. I managed projects for:
Changi Airport Group — permit and schedules management systems for one of the world's busiest airports. The kind of system where downtime isn't an inconvenience, it's an operational crisis.
Singapore Red Cross — volunteer management and donor management systems. Two interconnected platforms that had to handle sensitive personal data, complex matching logic, and reporting requirements that changed with every campaign.
Singapore Management University — management systems for academic operations.
Enterprise systems taught me that the hardest part isn't building the system — it's understanding the organisation well enough to build the right system. Every client had processes that existed for reasons nobody could fully articulate. The PM's job was to excavate those reasons before deciding what to keep and what to change.
CommonTown: EdTech at scale
CommonTown was different — an A*STAR Singapore spin-off with deep R&D capability, specialising in education technology. This wasn't client services; this was product management.
CommonAcademy
The flagship product I managed was CommonAcademy — a full-featured learning management system designed in collaboration with master educators.
The platform was comprehensive:
- Course design with multi-level units, drag-and-drop course maps, and real-time preview switching between designer and learner views
- 20+ question types — multiple choice, fill-in-the-blank, cloze passages, matching, drag-and-drop, image hotspots, audio and video question sets
- Learning games for vocabulary and concept reinforcement
- Collaborative tools — mindmaps, shared canvas, threaded discussion forums
- Language learning specialisation — speech technologies including Text-to-Speech and Speech Evaluation in English, Chinese, and Arabic
What made CommonAcademy different from typical LMS platforms were two features:
The Learning Flow Engine — an adaptive learning system where course designers could define conditional branching rules through a visual interface. Based on quiz performance and number of attempts, learners received differentiated content paths. This wasn't simple pass/fail routing — it was genuinely personalised learning at scale.
Deliberate Practice — an application using Item Response Theory (IRT) algorithms to compute learner ability in real time. It generated adaptive quizzes, factored in the natural tendency to forget over time, and reintroduced older material for spaced repetition. Teachers could track progress through radar charts and topic-level heatmaps.
The platform was deployed to schools across Southeast Asia and Hong Kong.
AI writing tools
Under my leadership, we added AI-powered features to CommonAcademy: Writing Coach and Writing Feedback. These were early — before ChatGPT made AI writing tools mainstream — and they taught me something important about AI product design: the technology is the easy part; the pedagogical framing is what determines whether teachers adopt it or reject it.
Qoqolo
I also managed Qoqolo, a Student Management System deployed across Star Learners Preschools — approximately 45 branches across Singapore. Different product, different audience, same discipline: understanding institutional workflows before designing software to support them.
What I learned
Complexity is not the enemy. Early in my career, I thought good PM work meant simplifying things. This chapter taught me that some problems are genuinely complex — multi-stakeholder, multi-country, multi-curriculum — and the skill isn't simplification, it's holding the complexity while making it navigable for everyone involved.
AI features need context, not just capability. The Writing Coach worked because we designed it within a pedagogical framework that teachers already understood. It wasn't "AI that writes for students" — it was "a tool that helps students improve their writing through structured feedback." The framing made the difference between adoption and resistance.
Enterprise and product are complementary lenses. Managing enterprise projects at WeesWares taught me to listen to organisations. Managing products at CommonTown taught me to listen to markets. Both are forms of the same skill — understanding what people need before they can articulate it — but they exercise different muscles. Having both made me a better PM than having either alone.
I earned my PMP and PSM certifications during this chapter — formalising the project management practice I'd been building since Chapter 02.