Conversations with Li Hongyi

By Joyce Er and Mock Yi Jun

Li Hongyi is Deputy Director (Data Science & Artificial Intelligence Division) at GovTech. He studied economics and computer science at Massachusetts Institute of Technology (MIT) before a quick stint working at Google, and thereafter entered the public service. He has been part of the teams working on,, and On his commute to and from work, he takes the opportunity to listen to audiobooks on a whole range of topics from physics to psychology to computer security.

HY: There are five main tranches to my work: developing products, getting internal people on board, getting users on board, project management and talking to your people one on one to make sure they’re developing.

The core work is figuring out how to develop products. This involves everything that goes with software engineering and product design – talking to users, figuring out user interface design, figuring out software architecture, and the actual writing of the code.

The second part is working the bureaucracy. The government has to account for public monies so there are processes to clear before you can push out a product. If it were a normal tech company, you could build a product and say “Here you go” and it would be pushed out. It’s not like that here. You’ve got to clear internal processes, and get people on board – and because not everyone has the same understanding of what is good, you need to spend some time convincing people of your idea. This is true in any organisation – especially so here.

The third part is figuring out your user strategy. With the product and internal stakeholders settled, now you need to get your external stakeholders on board. For, it’s about identifying your key user groups and figuring out how to engage them. For, it’s about figuring out who the people who care about data are in Singapore and reaching out to them to get them to use your product. For, it’s about trying to identify users in government agencies. A product – no matter how well-built – if not used, is nonexistent.

These are the three main chunks of work necessary to get a product out to the public, and then you have running the team itself, which involves two main things. The first is basic project management: keeping track of what needs to be done, keeping everyone informed of what’s happening, and trying not to be too heavy-handed. The next is making sure that the interests of the people on your team are being taken care of, that their careers are being developed, and that they don’t feel blocked in their work.

HY: Something people wouldn’t typically hear about is that getting agency buy-in was by far the hardest part of the project. The initial prototype for the app was built in two days. The alpha test lasted a month, and the beta test four. However, the hardest part was getting that first go-ahead for the pilot and managing all the cooks in the kitchen.

When I first joined the government, my mental model was that, as a technical guy who knew how to build products, I could come here and build good products, which the government would use since the government was looking for technical capabilities to build products. But that didn’t happen. Instead, you come here, you build something useful, and you realise that the government makes certain choices which work against such products. We lack products not because we lack resources but because we choose to – which means that you need to work to change that choice. You need to figure out who is making these choices and why they are making these choices. They often have reasons – not terrible ones, but nonetheless if there is a better way, then you need to figure how to convince them of it.

For, the Urban Redevelopment Authority (URA) and the Housing Development Board (HDB) had a similar idea they tried out years ago, which used phone texts. They knew it was a good idea. This isn’t new; many have had this idea before. But URA and HDB found it was too expensive. Basically, they were paying full price per text sent out, which is a very expensive system for digital parking. So, they didn’t even want to try it. We needed to demonstrate that it could be more efficient and cost effective. Funnily enough, we run so cheaply they thought we were making things up. runs on about $700 per month of infrastructure, not counting all the other stuff, of course. For perspective, a single server in the Singapore government typically costs about $5000, depending on the server.

That’s something you have to deal with: figuring out how to convince people that the horrid way things are isn’t the horrid way things have to be. It takes persistence; we talked to URA and HDB for at least six months – making slides, presentations, and addressing concerns. If it wasn’t for sheer pigheadedness, we would’ve given up a long time ago. But we didn’t and we got there eventually.

I think we need to overcome our choice to be afraid and stick to what we have even though there’s something better. If not, we will lack innovation not because of lack of ability or lack of resources, but because we choose not to innovate. And figuring out how to change that choice is key.

HY: Everyone’s heard how Google is a wonderful place to work. So, yes, it’s a good place to work: you get paid very well, you get lots of benefits and free massages, and you work with some of the best people on products that are used by hundreds of millions of people around the world.

The thing to realise about Google and most private companies is that, even if you are incredibly successful – you’ll see that the bunch of people working there are incredibly smart; some were valedictorians of schools and everyone said they would be successful – is that you would have spent the last year or two of your working life figuring out what colour advertisements should be to make ten million more dollars in revenue. It’s big money, but is the world a better place because the ads are now a slightly bluer shade of blue? Probably not.

In the government, if you get a thing done – like fixing parking, which some might say is rather a first world problem we have – and remove that pain, you make people happy because their lives are better. Our goal is not only to fix these small things, but to fix the important stuff. If something small can already bring such joy, what more with fixing the more important stuff?

That’s the difference. In government, you are going to work on the most important problems.  It’s a question of what you want to do.

HY: Oh, tremendously. When I first went to MIT, I was there to study economics as a major. I later wanted to do something else together with economics, because I felt it was good to have more exposure. I had taken classes in computer science and architecture, and I ultimately decided on computer science. Being at MIT, around a whole bunch of engineers did somewhat influence me to take up computer science – which is obviously very relevant to what I do now.

I remember that, in my junior year, a friend encouraged me to apply for this Google internship. I’d just taken a couple of computer science classes and didn’t even have a resume – I hadn’t prepared one because I already had a job back home due to my scholarship. But my friend pushed me to do it. I tried it out, got the job, and did an internship there for a summer. The biggest thing I realised from that was that there are so many things that you could do that aren’t even that hard, sometimes it’s just about taking that step of trying.

I think the biggest difference between students at MIT versus other universities is not that students at MIT are any smarter. If you did well in the education system here, you could do well at MIT. The difference from the people I’ve met there is not that they were inherently smarter, but that they were willing to try harder. They weren’t just content with graduating and getting a good grade or class of honours. They would go and try to build a solar car or start a clinic to treat heart attack patients in some country. They weren’t inherently good at these things – because nobody inherently has such knowledge. But because they were willing to try, even though they didn’t know anything, they would get good at it after a short period of time figuring out the basics.

When you’re a student, there’s this mystique built around having a long professional career, and having experience. But having seen how many organisations and teams are run, if you can put together a school fun fair, for example, that is the same set of skills that you need to run an organisation. I’m not even kidding. It’s just making sure that you’re keeping track of what’s going on, ensuring that your team knows what to do, communicating openly and honestly, prioritising your goals. It’s the same thing.

I had a friend who was just inherently super organised. We were planning a trip to Costa Rica for our group of friends. She, for that trip, had put together a binder of arrival times, pictures of restaurants and contact numbers. And she did that because that was just the kind of person she was. None of that is inherently a super high intelligence task, but that one binder is more organised than some multi-hundred million-dollar projects I’ve seen, and this was for our little group trip to Costa Rica. If you can do that, then I think succeeding professionally isn’t as far away as you think. It’s not as mystical as you think. I think that was a big insight.

HY: The culture of the team focuses on two things. One is figuring out what needs to be done, two is taking people seriously. Treat people like adults; hold them accountable for the work they do and their performance. If you don’t trust them to do basic things, then you can’t do much with them. Try to develop them; it’s a big part of the team culture is. I have documents to keep track of people’s development, because that’s another important thing that people tend to ignore. For every person, I ask them to keep track of their goals, whether it’s developing as a software engineer or as a project manager, and then ensure that at least some of the time people are working on things that they care about. And you need to actively manage your people and put them on projects they care about. Your job is to develop your people and if you only focus on the projects, you start sacrificing your people.

We don’t have a fancy pantry; we don’t have free massages or gourmet snacks. But if you need to work from home because you have a doctor’s appointment or your kid is sick – yes, of course, your life is important – and recognising that helps the team.

HY: Oh yeah, all the time. Just recently I learned to touch-type properly. I never used all ten fingers like you’re supposed to – I typed weirdly growing up. For about two weeks, I set aside 15 minutes a day to practice touch typing, so now I can kind of type better than I could before.

I watch a lot of YouTube videos on various parts of tech: how to build things, frameworks, technologies. I actually find this very useful because every year or half a year in tech, the state-of-the-art way of building things changes, and if you don’t keep up to date, you just use old tech. It’s not that the old way doesn’t work, it’s just that with new tech you can do the same thing much more easily. And so it’s worth the time to keep up. Whenever we’re starting a project, instead of saying, “Oh yeah, we’ve built this before”, we just Google if there is an app to do what we need. So, if we’re building the first thing we do is search open source form builder, and you just see what people are doing. I’d say 99 times out of a 100, you won’t find exactly what you want, but you will find someone who has done 60 to 70 percent of the work for you. That’s why the apps we build are done so much faster, versus trying to charge into it blindly.

I listen to a lot of audiobooks. I try to specify the big broad categories of human knowledge and try to listen to an introductory book on each. It’s nowhere near a university education but at least you understand the big thought processes behind, for example, physics, biology, chemistry, history, economics, psychology, computer security and a few others, design, and things like that. You’re not going for becoming an expert, but just to and from work, you listen to these things, and over time you build a basic understanding of the thought processes in these fields and what the main things they think about are. You get a rough sense of what the big conundrums and focuses are. When you talk to someone about these things in the future, you at least know where to start to look things up and go deeper.

For example, I listened to the Selfish Gene recently, by Richard Dawkins. There’s stuff which may or may not be right, or which might or might not require additional investigation, but at least it gives a brief introduction. I listen to The Art of Deception by Kevin Mitnick, which is about social engineering and hacking in the 80s or 90s. These were the times when hacking involved picking up the phone and dialing the tone codes to get access to someone’s phone network. The specifics are probably obsolete now, but the general approach of figuring out how a hacker thinks and how people think about security, identifying threats and vectors of going in, is very interesting. Because they’re written as books, they keep your attention more than a textbook. So, I would say that’s one way I try to learn things.

HY: I think the biggest highlight of my job is that we have a very rare opportunity to really do something nobody in the world has ever seen – which is a truly modern, functioning government. It’s a weird thing to say, right, especially in Singapore.

But if you were to look at the world, you see a place where the private sector is increasingly using more sophisticated technology to do extremely trivial things. Like, if I’m buying soap online, there is an AI detection system to figure out where my payment came from, link it to what I’m going to buy next, recommend brands and figure out whether I’m likely to cheat and all that.

There are few countries in the world that have the opportunities we have. The Americans and British are having political trouble, and the Europeans and the South Koreans. You look around in the world and you see people fumbling and scrambling to be the next guy in charge while the country stands still. Singapore has a lot of problems, but one of the things we don’t have is a lack of political stability and a lack of resources. For all our issues, we are in the best position for a few people to do something really right.

That, I think, is the exciting part, where you feel on our team that, if you are really successful at your job, not only does Singapore develop and not only do we alleviate suffering and reduce inefficiency, but then we become this example in the world of what a country run right can look like. And people will come and look at us, and they will see what society can be like. Even if they have always felt that this was possible, by seeing with their own eyes that it can happen, it becomes that much more powerful. It’s no longer just this theoretical dream, they see that it’s been done and so they can do it too. And hopefully that pushes other countries to do better, and we move towards a better world because of it.

HY: For people who are interested in applying, the things we look for are straightforward. First, we look for ability. Simply, are you good at what you do? If you’re working in tech, you need to understand tech. If you are working in UI design, you need to understand UI design. It’s not someone else’s job, it’s yours. You need to understand your field even if you’re administrating someone else to do it. Next, we look for initiative. Our job is to push the country to change, and so if I have to push you too, that’s not very useful. Third is communication. Can you work with us, can you get along with the team?

Finally, the most important bit is the values. I’m not interested in hiring people who, if I tell them to build a doughnut-finding application for the government, will do it. Because we have plenty of people in government who are capable, but who, over the years, have just gotten beaten down and do what they’re told. I need people who want to do the right thing. Because if you do what you’re told and you don’t have a strong set of values, at some point you’re going to have a bad boss, and they’re going to tell you to do something dumb, and those bad ideas gain momentum real quick. Almost by definition bad ideas have fewer constraints than good ideas, so they gain momentum faster. I look out very acutely for people who, in their decision making process, do their job not because it pays well or because it’s a good career. We take people seriously. Ultimately I want people who want to do this job because they think that there is more to life than what we have now, and we can make a difference there.