Software Engineer - A26268

About this role

Activate Interactive Pte Ltd (“Activate”) is a leading technology consultancy headquartered in Singapore with a presence in Malaysia and Indonesia. Our clients are empowered with quality, cost-effective, and impactful end-to-end application development, like mobile and web applications, and cloud technology that remove technology roadblocks and increase their business efficiency. We believe in positively impacting the lives of people around us and the environment we live in through the use of technology. Hence, we are committed to providing a conducive environment for all employees to realise their full potential, who in turn have the opportunity to continuously drive innovation. We are searching for our next team members to join our growing team. If you love the idea of being part of a growing company with exciting prospects in mobile and web technologies that create positive impact on people’s lives, then we would love to hear from you. Co-Development Business Unit is looking for a Software Engineer This is a fixed term contract role. The engagement is 1 year. Internal Code: A26268 We are looking for a Full-Stack Software Engineer to join the Singapore Examinations and Assessment Board (SEAB) team, working on Scanvas — SEAB's next-generation AI-powered platform for digitising national examination scripts. This is not a role where you spend your days inside a browser tab. You will write web applications, yes — but you will also write code that talks to scanners, ships as a desktop application to scan-room operators, and integrates with AI services that process hundreds of thousands of exam pages under real time pressure. If that breadth excites you rather than unsettles you, read on. The Singapore Examinations and Assessment Board designs and administers national examinations — PSLE, GCE N/O/A-Levels, and more — that shape the educational journey of hundreds of thousands of students every year. Trust, integrity, and precision are not aspirational values here; they are operational requirements. SEAB is in the middle of a technology modernisation. Scanvas replaces existing scanning stack with a platform that SEAB owns, controls, and can evolve — and it must be production-ready for national exam seasons where failure is not an option. What Scanvas Is Scanvas is an Optical Mark Recognition (OMR) and script digitisation platform. It does two things: On the hardware side: scan-room operators use a desktop application (ScanController) to drive high-throughput TWAIN-compatible document scanners, validate scans in real time, and upload captured images to the cloud. On the platform side: a web application lets assessment staff generate script templates from exam paper PDFs using AI (zone detection, LLM-assisted layout), process uploaded scans against those templates, and route digitised responses to the marking system. The two sides are tightly coupled. Understanding one helps you build the other You will touch real hardware. The scanners are physical machines in a physical room. When something breaks — a driver handshake, a paper jam mid-batch, a scan that arrives corrupted — your code needs to handle it gracefully, and you need to be comfortable debugging it. The desktop application runs on government-managed Windows machines, not your MacBook. Distribution, update management, and compatibility constraints are real engineering problems here. AI is not a feature you bolt on. Template generation, zone detection, and handwriting OCR are core to what Scanvas does. You will integrate and tune these pipelines, not just call an API and hope. The workload is lumpy and extreme. 12 million pages must be scanned and processed across a three-month window each year, with days that peak above a million pages. The system you build will either hold or collapse under that load — there is no gradual rollout. Build and maintain the Scanvas web platform — API routes, background job queues, database models, and the canvas-based template editor that assessment staff use to configure OMR zones. Build and maintain the ScanController desktop application — the interface between scan-room operators, the scanner hardware, and the cloud backend. Integrate hardware: scanner SDKs, driver interfaces (TWAIN/ISIS/WIA or vendor APIs), and scan-capture pipelines that validate quality and handle errors before upload. Integrate AI services: LLM APIs for OMR template generation, computer vision for zone detection, and document OCR/ICR for handwritten and typed responses on mix-form booklets. Design and optimise for throughput — job queues, concurrency limits, rate-limited external AI calls, and S3 storage pipelines that must sustain peak exam-season load. Collaborate with the Product Manager and operations team to turn real scan-room workflows into software that people trust under pressure. Maintain engineering rigour: TDD, typed code, clean API boundaries, meaningful CI gates, and documentation that the next engineer can actually use Requirements What are we looking for? The fundamentals matter most. We are not prescriptive about your current stack. Engineers who understand why systems behave the way they do pick up new frameworks quickly. Engineers who only know their current framework struggle when the problem doesn't fit the tutorial. What the fundamentals look like in practice: You can design a system given a set of constraints and articulate the trade-offs you made. You don't need a framework to do this. You understand concurrency — threads, event loops, async/await — well enough to reason about race conditions and bottlenecks, not just write await and hope. You write tests before you're asked to, not because it's policy but because you've been burned by not doing it. You've debugged something that crossed a system boundary — a network call, an OS driver, a hardware device — and know how to work methodically when the usual tools don't apply. What you bring: Solid full-stack experience — web frontend, backend APIs, relational d

View the original posting ↗