The strongest overall match: hands-on design of AI assistants and platforms, with ownership from discovery through shipping.
GovTech is the lead agency driving Singapore’s Smart Nation initiatives and public sector digital transformation. GovTech develops the Singapore Government’s capabilities in Data Science and Artificial Intelligence, Application Development, Smart City Technology, Digital Infrastructure, and Cybersecurity.
AI is rapidly transforming how governments deliver services and how public officers work.
At GovTech’s AI Programme, we build the platforms and experiences that enable Singapore’s public service to safely and effectively adopt AI at scale. As a Designer, you will help shape the next generation of AI-powered experiences used across government.
See moreFound in the 30 July 2026 job search and not yet reviewed.
AI product builder and designer with six years translating complex products and technical ideas into clear experiences, practical tools, and reusable workflows.
GovTech builds and operates shared digital products for Singapore’s government, funded as a public agency and used by public officers, agencies, citizens, and businesses. The AI Programme turns foundation models into secure assistants, developer platforms, evaluation tools, and reusable interaction patterns for government work.
GovTech began moving from vendor-managed projects to continuously owned products, so this role is likely expected to influence and improve a live product rather than hand off screens.
GovTech announced an AI-agent registry and oversight layer for 150,000 officers, making agent control, transparency and recovery live design problems.
Pair reached 90,490 monthly active users and reported 86 minutes saved per task, so the next design question is quality and trust at scale rather than basic adoption.
Deep Microsoft-workflow integration with enterprise data protection.
Broad model capability, agents and tools with managed enterprise controls.
AI embedded across Workspace with existing tenant security.
GovTech’s durable edge is the government-specific product, security and governance layer around the models, not privileged access to a better base model.
I am a designer who turns complicated work into something people can actually use. Over six years I have researched unfamiliar products, found the clearest path through them, and built both the experience and the reusable system behind it; more recently I have moved that practice directly into AI products through Pitchcraft, local-LLM and agentic prototypes, and 12 hours of prompt-engineering curriculum. GovTech’s AI Programme is compelling because the design problem is not putting a chat box on a workflow — it is making powerful, imperfect technology understandable, trustworthy, and useful across many public-service contexts.
AngelHack proposal workflow. Proposal work was repetitive, but the useful input lived in messy meeting information. I designed an AI workflow around that real input and reduced proposal creation time by 90%.
AngelHack enablement system. I turned one-off decks into a shared structure of templates, sales decks and more than 60 case studies. The system supported more than $10 million in sales pipeline.
Pitchcraft. I translated tacit pitch-advisory judgement into the prompts, feedback structure and interface of a working AI web app. It draws on six years of work, including decks that helped one client grow from $10 million to $20 million before acquisition.
SMU and Tribe curriculum. I developed 12 hours of prompt-engineering content for business learners, turning model behaviour into exercises and reusable mental models. Learners could test, observe and improve outputs rather than depend on magic-language claims.
Local-LLM finance tool. I built a fully local Python app that turns PDF card statements into structured transactions. Privacy shaped the architecture and inconsistent statements shaped fallback behaviour, converting a manual task into a repeatable workflow.
AngelHack website overhaul. I translated complex innovation work into a clearer journey for enterprise buyers and led the experience overhaul. The site brought in more than $2 million in inbound opportunities.
AngelHack automation adoption. I built more than 15 automations, but the hard part was fitting them into handoffs people would keep using. Alongside custom GPTs, they turned tasks that took hours into tasks that took minutes.
Answer: The title is less direct than the work. Research, interaction and information design, prototyping, AI workflow design and hands-on building are the through-line — let me show you one product example.
Answer: Lead the case study with what you observed, the constraint it changed, the prototype you used to test it, and what happened next. Do not defend the portfolio in the abstract.
Answer: I have not. What I bring is practice turning tacit judgement into reusable systems, designing for non-specialists, building around privacy and adoption, and learning technical domains quickly.
Answer: I would identify affected users and harm, define human review and recovery paths, expose limitations at the decision point, and test failure states before polishing the happy path.
What is the hardest design problem the AI Programme needs this person to own in the first six months??It reveals the real product, team and seniority expectation.
You are moving toward continuous product ownership. What has that changed in practice for how design, product and engineering make decisions together??Grounded in the July 2026 operating-model shift.
Across Pair, AIBots and the newer agent work, which interaction patterns are mature enough to standardise, and which are still being discovered??Directly answers the posting’s systems remit.
How do you decide when an AI experience has earned a user’s trust, rather than merely attracted usage??Pair already has scale, so this moves beyond adoption metrics.
Where does a designer have authority to challenge a policy or technical constraint when it creates a poor user experience??The answer changes whether this role offers real product influence.
Could you help me understand whether this is a permanent GovTech role or a fixed-term or vendor contract, and how this team is affected by the current transition??Ask calmly near the end; the restructuring makes the terms material.
GovTech cut 93 roles in July 2026, and says 7% to 9% of roles will be affected over two years, even though it expects total employment to grow after the transition.
Glassdoor shows 3.5 / 5 overall from 908 reviews but 2.9 for senior management; Indeed’s much smaller sample also rates management below pay and flexibility.
The posting says full-time, but the coffee-chat invitation and database record do not establish whether this is permanent, fixed-term, or vendor-contracted. Close this in the conversation.