Volunteer Spotlight: Billy Zhao
This month’s spotlit star is Billy Zhao. Billy is currently a senior data scientist at multinational education company Education First. He has also been busy launching a hate crime support service called On Your Side within the UK for East and Southeast Asian communities. In his free time, his default place to volunteer is DataKind UK.
Tell us a bit about your data background
I started in the field of data science three or four years ago. I was working in a startup called AI For Good UK, working on chatbot analytics: one to do with domestic violence education and signposting, the other with sexual health education for young women. It was really interesting work in the data for social good space. That’s around the time I started volunteering for DataKind UK. I was in London and went to a few DataDive weekends, and then became a Data Ambassador.
After that I worked for the Alan Turing Institute on a project finding different clues within public data to infer corruption signals in the public office — very interesting as well. For the past two years I’ve been working with education company Education First on a wide range of data applications, and focusing a lot of energy on the marketing and sales applications of machine learning.
What’s your daily work at the moment?
In terms of tools, I mainly use Python.
My daily work is with a wide range of businesses. One part of Education First is teaching languages, and another is travel. Recently I’ve been working to try to understand how different online or offline marketing investments will come together, in order to measure the impact and be more efficient with our marketing spending.
How did you get into data originally?
My path is probably quite traditional: learning data science and machine learning in university. I did an engineering degree, and then after a couple of years I realised I’m not cut out for bridges and trains! I decided that maths, probability, and programming sounded fun. I started learning programming by myself because the school didn’t really teach that, and I was hooked.
The main thing that attracted me to AI and machine learning within engineering is that this field is so new, and it’s ever-evolving. In the conversation we keep having around data decision making, it’s a very new and trendy thing, and I got attracted to it and working in roles related to data.
What would have been the best thing to know when you started learning programming?
I think I can always learn a bit more! When you hit a roadblock it can be really devastating, you kinda lose your motivation. What I did back then, that I would advise other people to do, is use projects to learn programming, because doing a project is the best source of motivation you can give yourself. If I try to follow some course it never works out: there’s no ‘why’. For me, the motivating bit is ‘if I join this project now I have to deliver this’. The desire to learn is there because you’re solving a pretty cool problem, and second there’s the deadline — if you don’t get on it, you will delay your project.
I think solo projects have less accountability, but on a group project, others will be asking for the output. I’d still advise people that that’s the best way to learn. When I was doing a lot of data analysis, data dashboarding, and visualisation by myself, by coming to a DataDive weekend I still learned tricks working with other volunteers. The DataDive project was really cool in terms of working with a huge range of people. I think I even talked about my day job and people were very keen to give me some direction and advice, so it was really good.
Tell us about a data project that inspires you
One great project I mentioned was with the Alan Turing Institute. I was collaborating with people from a wide variety of backgrounds: North America, Mexico, Colombia, Germany, a very diverse international team. I think it started my interest in looking at public data in the public space, and holding politicians accountable.
The project was looking at public data to see if you can infer any corruption within public offices. Usually I’m the guy who knows techniques, but I have no subject matter expertise. So we were talking to a lot of political scientists about how different signals could infer corruption.
It was really fun to go through that whole process, to understand that you have to rely on subject matter a lot. And then the challenge is how to encode that knowledge into your algorithm. It was really nice to collaborate with someone quite senior in their position. I learned quite a lot from that project, other than it being really impactful in its nature. It was a proof of concept where we built a lot of pipelines, which led into a call for papers at the World Bank. It wasn’t applied in practice, but it was feeding into further research so we were happy to see that.
Did you always think about going into data? If you weren’t doing it, what might you be doing?
If I weren’t doing it I might become a consultant of some sort, because I love solving problems. I think part of being a data scientist in a company is about communicating upwards and vertically to your peers about the value of data science and how it can help.
In an AI or tech company you might be building an AI product already, so it’s quite easy and justifiable for you to work on something, because maybe the CTO or the tech lead asked you to do it.
But in a company that’s a bit more traditional, or not data-sciencey, it’s about problem solving. You have to find out what the problem is, you have to find out the best solution. And you have to work through how this is going to affect all the people who are involved, making sure they don’t panic, and know how it fits into their day-to-day lives. I really enjoy that component of the data science job. So I might be a strategy consultant because it’s fun talking to people, finding out what’s wrong, and thinking of a way you can fix it. Making something tangible that you can use and apply.
That’s the power of it — I’ve been exposed to many industries. Using data decision-making is so widespread that you can be in any industry. You quickly ramp up on what’s in a domain, and use the knowledge of domain experts. I’m quite domain agostic. My current role is mainly marketing, but prior to that I had no knowledge of marketing whatsoever. You will quickly learn everything you need to learn to make data science work. I think that’s the beauty of data and data science for me — you can pretty much work in any industry.
How did you find DataKind UK and why did you decide to get involved?
I was quite proactive around the time I first got to London. I was really keen, because I was working in a startup and was the only data scientist there. So it was crucial for me to find a community, meet people, and learn from others. I liked volunteering at university and even before that, I was interested in international development.
I was always thinking, how can I combine data with volunteering? I was searching all kinds of meetups to go to. I remember the first day that I actually met some other people that were of a ‘data for social good’ nature and was like ‘oh right, you guys all hang here’, DataKind UK seems pretty fun. So I just hopped on.
What surprised you about volunteering with us?
I think what surprised me is that you are able to make this work! I know other people that are trying to start something similar within their company, but you are becoming the hub where different charities come together. DataKind UK is able to deliver something tangible so that charities keep coming back. It’s surprising and very inspiring!
You’ve been a Data Ambassador and led DataDive projects — why did you get involved?
I guess the main motivation was to become more involved with it and to see ‘under the hood’. It was interesting to see what data they have and then really shape those questions. I actually enjoy that even more than diving into the data during the weekend! Answering the question, scoping it, it’s very fun.
I found the exchanges with the charity representatives most fulfilling. By having a meaningful two way conversation, I was able to understand the data challenges in the charity sector more clearly, and I was able to share some of the best practices around data in the private sector, which I hope the charity can slowly adopt.
Do you have any advice for future Data Ambassadors?
Enjoy the process! I think it’s really nice working closely with a charity compared to a DataDive weekend where you show up and do analysis. You can forge a better, deeper relationship and really understand a lot more of their operation, their struggles, their concerns. Things you can observe for maybe working in the nonprofit sector in the future.
Is there a resource you’d recommend to the community?
I stopped reading a lot of newsletters, but nowadays I’ve found when I do housework and I’m feeling like I want to study, I’ll plug into some lectures. On YouTube you can find the Stanford MLSys Seminars.
Everyone was talking about this new concept MLOps, and before watching the seminar I didn’t really know what it was about — it really educated me, and triggered my interest in that area. It was a great seminar and I’d recommend it!
Tell us something completely non-data related about yourself!
I play a lot of basketball with a local team not that far from where I live, called London Phoenix. It’s an amateur London league — I think between amateur and professional there’s a metropolitan league, but we sortof fell out of that, and amateur is not as intense. In the metropolitan league I think you do a lot of travelling, but we play in a fixed place every Sunday, so I’m happy with that!
If you’re curious about becoming a volunteer with DataKind UK, please take a look at our volunteering page!
