Senior/ Software Engineer, AI Programme

GovTech is the lead agency driving Singapore’s Smart Nation initiatives and public sector digital transformation. As the Centre of Excellence for Infocomm Technology and Smart Systems (ICT & SS), GovTech develops the Singapore Government’s capabilities in Data Science & Artificial Intelligence, Application Development, Smart City Technology, Digital Infrastructure, and Cybersecurity.

At GovTech, we offer you a purposeful career to make lives better where we empower our people to master their craft through robust learning and development opportunities all year round.

Play a part in Singapore’s vision to build a Smart Nation and embark on your meaningful journey to build tech for public good. Join us to advance our mission and shape your future with us today!

Learn more about GovTech at tech.gov.sg.

[What you will be working on]

The AI Programme builds shared AI infrastructure that agencies can run in production, instead of rebuilding the same foundations independently. It sits within Government Digital Products, and it owns a small number of central products that a lot of other people's work now depends on.

Three of them shape the job. There's the agent surface public officers use to get through the day: it reaches into their mail, their documents and their case systems, so it has to be quick, honest about what it just did on their behalf, and safe to point at real records.

Underneath sits the developer platform, where teams across government bring their own agents, MCP servers and skills and move them into production through review, approval and audit.

And there's the model and capability layer that most agencies meet first: a compliant gateway to multiple production-approved models behind an OpenAI-compatible API, alongside shared speech, vision and document capabilities (live transcription, image and video classification, key information extraction) that every agency would otherwise go and procure separately, with guardrails, quotas and cost attribution already attached.

Building an agent demo is easy now, and that's fine, because it means the hard parts have moved. Most of what we work on lives after the demo. An agent decides to call a tool that sends an email on someone's behalf, and the approval gate for that has to be enforced by the runtime, not requested politely in a system prompt. A tool loop needs a stop condition that doesn't depend on the model deciding it's done.

Token spend has to trace back to the team that caused it, and when an agency's agent gets slower this week, someone has to be able to say why. Rollback gets interesting when the artefact is a prompt, a tool schema and a model version rather than a binary. Very little of that is model work. It's orchestration, platform and operations, which is roughly the job.

You'll work across all of it: the surfaces officers see, the runtime underneath, the platform other engineers build on, and the unglamorous business of keeping the whole thing up. Your user is a public officer with something to finish today. Your customer is often another engineering team in another agency.

What you'll do

• Build and run the surfaces officers use daily. Streaming, approval interrupts, resumable sessions, and interfaces that stay readable when an agent is doing four things at once
• Build the runtime beneath them: tool orchestration, approval gates enforced in code, loop limits, per-turn tool filtering, memory and checkpointing, model routing and fallback
• Work on the platform half, where the job is to let other teams onboard their own agents, MCP servers, skills and models, with the policy checks and promotion workflows that let them move to production safely and repeatedly
• Extend the shared capability APIs (speech, vision, document extraction) and keep them fast and cheap enough that agencies reach for them instead of building their own
• Integrate with what government already runs: identity, mail and calendar, document stores, agency line-of-business systems, the existing CI/CD pipelines
• Instrument the things you'll want to argue about later. Tool call traces, latency, error taxonomies, cost per team and per feature
• Build evaluation into delivery, so changing a prompt, a tool or a model becomes a measurable decision instead of an argument
• Run it. Deployments, incidents, capacity, cost, and the support queue of officers who found the case nobody thought of
• Work with agency teams as users and as builders, and turn the third bespoke request into a platform feature
• Take part in security and privacy reviews on the assumption that these systems get audited, because they do

You'll thrive here if

• You want the thing you wrote to be used by someone doing real work the next morning
• You're interested in what happens after launch, not only up to it
• The constraints (classification, audit trails, approvals, cost accountability) read to you as design problems rather than obstacles to route around
• You have a view on what belongs in a prompt and what belongs in the backend, and you'll defend it
• You'd rather ship something narrow and run it properly than demo something broad
• You use agentic coding tools every day and have opinions about where they help and where they quietly cost you
• You can talk to a policy officer in the morning and debug a streaming protocol after lunch
• You expect the model layer to change under you every few months, and you design so that's survivable

[What we are looking for]

We're deliberately open on background. This is a broad role, and the strongest signal is something you built, shipped and then had to keep running, not a number of years.

• Strong engineering fundamentals and fluency in TypeScript or Python (we mostly write React and Node on the front, Python and Node behind it, Postgres, AWS)
• Comfort moving across frontend, backend, data and cloud rather than settling into one layer
• Something you've taken end to end: designed it, built it, deployed it, watched it break, fixed it
• Working knowledge of containers, CI/CD and infrastructure as code (Terraform or similar), or the appetite to pick them up quickly
• Understanding of authentication, authorisation and careful data handling, which matter more than usual when an agent is the one holding the credentials
• Practical experience with LLM APIs, tool calling, retrieval or agentic workflows is welcome. Strong engineering and real curiosity will beat an AI-heavy CV with nothing shipped behind it
• Fluency with agentic development tools such as Claude Code, Codex or Cursor
• Writing you'd be happy to put in front of someone who will never read the code
• Reliability with data that belongs to the public

Useful, but genuinely not expected: evaluation harnesses for AI systems, MCP server authoring, streaming protocols (SSE, WebSockets), managed agent runtimes on AWS or GCP, Kubernetes, observability stacks, or any prior public sector work.

If you've built one substantial thing you're proud of and can walk us through the decisions inside it, we'd like to talk. You don't need to have worked on AI systems before.

What you can expect

• Users from the first week, across agencies with genuinely different needs and maturity
• An unusual split, where you build the platform and also build on it, so you live with your own abstractions
• Ownership from design through operations, and the autonomy that has to come with it
• Work on production problems in agent systems: tool orchestration, governance, memory, evaluation and cost control under real audit and reliability constraints
• Colleagues across GovTech's AI Practice, Government Technology Office and product teams, and a mandate broad enough that good work travels
• Support to stay hands-on: learning budget, conferences, and time to go deep on the parts of this that nobody has settled yet

What we offer you

GovTech is an equal opportunity employer committed to fostering an inclusive workplace that values diverse voices and perspectives, as we believe that diversity is the foundation to innovation.

Our employee benefits are based on a total rewards approach, offering a holistic and market-competitive suite of perks. These include leave benefits to meet your work-life needs and employee wellness programs.

We champion flexible work arrangements (subject to your job role) and trust that you will manage your own time to deliver your best, wherever you are, and whatever works best for you.

Learn more about life inside GovTech at go.gov.sg/GovTechCareers.

Stay connected with us on social media at go.gov.sg/ConnectWithGovTech