Anna Ying, Data Scientist, Government Digital Transformation, Government Technology Agency (GovTech) of Singapore

1) What does public service mean to you? Can you share more about your role in the public sector?

Public service, to me, is work that quietly improves daily life for people you will mostly never meet – rarely fast or glamorous, but it lands where it matters, and the groundwork compounds.

I am a data scientist with Government Technology Agency (GovTech) of Singapore, deployed to the Ministry of Education Singapore, working at two ends of one problem: building artificial intelligence (AI), and helping shape how it’s governed.

On one end, I helped build LangBuddy, a chatbot that helps students practise speaking their mother tongue, from features to safety evaluations.

On the other, I work on AI governance more broadly: it’s more than just a set of rules (that many people believe nobody reads); rather it’s the erecting of guardrails to make sure a chatbot is safe for use before a 14-year-old leans on it for advice or perspective at 10 pm.

2) Tell us about a project you championed. What impact did it have on the community?

One project I am proud of is LangBuddy’s evaluation work.

LangBuddy is meant to be a safe space for students to practise speaking, which means the bar cannot just be set at “a chatbot that just replies”.

We need to know whether the chatbot is safe, whether it is hallucinating, and whether it follows the pedagogical rules we set – for example, responding directly instead of giving a beautifully irrelevant answer.

I have been leading the work on refining its methodology: defining what good looks like, reviewing pilot conversations after use, and starting to design evaluation datasets and LLM-as-a-Judge rubrics so we can test changes more consistently.

Without this, testing becomes an endless loop of tweaking prompts and squinting to see if the model actually improved—or just behaved itself that day.

3) As a young professional, how has your unique background or perspective allowed you to identify a solution that others in your organisation might have overlooked?

For me it comes down to angle.

I came to AI governance as a data scientist, rather than from policy development.

I am thus newer to this, and step in without fixed assumptions about how it should work.

I will notice things from the build side: where a well-meaning rule wouldn’t survive contact with how a model behaves, or where governance is treated as paperwork, not something you can engineer into the process.

But the bigger factor is the environment – GovTech and MOE are places where a junior technical person can suggest something, be taken seriously, and handed the work.

That’s how I ended up here.

4) What is your personal strategy for maintaining your creative energy when faced with bureaucracy?

Funny thing – I am part of the bureaucracy now, at least the AI-governance bit, so I see it from the other side.

Mostly I get why it is there: much of it exists to hold a line on standards and quality, especially anything citizen-facing.

It still adds up to a long line of checks and approvals, and understanding why does not stop me sighing at it.

What keeps me unstuck is the environment – agile enough that when part of it is clunky, I can go and try to make it better.

5) If you had just one area to invest in to accelerate transformation in the public sector (regulation, technology, talent, etc.), which one would you choose and why?

Talent – people who can work across the technical and policy worlds, to be specific.

So much friction in public-sector transformation sits in that gap, and the people who can stand in there and translate from both sides go on to unblock obstacles.

We have got brilliant engineers and policy thinkers; the multiplier is anyone with a foot in both camps. If I could invest in one thing, it would be growing more of them – because that is what turns good intentions into things that ship.

6) What is your greatest ambition as you grow in your public service career?

What I want, more than any title, is to keep doing work that’s meaningful and genuinely useful – where what I’m good at meets what’s actually needed (there’s a Japanese idea, ikigai, that captures it).

I would also like what I build to keep standing on its own after I’ve moved on.

Right now that is AI governance in education – especially given how fast AI is moving into classrooms.

But I do not assume it will always look like this: what needs doing shifts, and so does what I am good at and drawn to – so I would rather move with it, to a new problem or team, than stay in one seat out of habit.

7) What is a “universal value” that connects everyone in your department – from interns to directors – and how do you use that to drive collaboration?

It is the habit of asking "so what?" – who’s this for, and what do they need?

The people I work best with, junior or senior, care that the work is good, and that’s what makes collaborating easy: when everyone’s optimising for whether it helps the person on the other end rather than credit, you pull in the same direction.

So when we get stuck, the fastest way through is that question – what does the person we are building for actually need?

8) What is the best piece of advice you’ve got for the next generation of public servants?

Pay attention to what could work better – the gaps nobody owns – and put your hand up to take one on.

That’s how I ended up in AI governance: I spotted a gap worth closing, said so, and asked to work on it rather than wait.

The one thing I would add, having learned it the hard way, is that you can’t take every gap – running yourself into the ground helps no one. Pick the one that matters most, and keep enough in the tank to do it well.

9) What is a myth you wish to debunk about young public servants?

There is a quiet assumption that a young, technical person who joins the public service cannot be doing anything interesting – because the serious tech, surely, happens somewhere else.

That’s the myth I would push back on. What I do is genuinely technical and rigorous – building AI, evaluating it, helping govern how it’s used safely at national scale. A different kind of hard problem from the one a startup’s solving, but just as real, and the stakes are public.

10) Write a letter to your future self in 2035.

Hey,

You’re 35 now, which feels so far from where I’m sitting tonight. I hope you still get annoyed at slow processes – the day that stops bothering you is probably the day you’ve stopped pushing, so stay impatient.

I hope you’re still following your ikigai – work that’s meaningful and worth doing, where what you’re good at meets what actually needs doing.

Maybe that’s still AI governance, maybe something you haven’t thought of yet. Whatever it is, I hope you chose it on purpose.

And on the hard days, try to remember why you started – not the title or recognition, but why it felt worth doing.

If something’s broken and wearing you down, don’t just go cold on it – be the change you wanted to see when you walked in.

Go easy on yourself – but not too easy.

Anna