top of page

WTF does a Data Scientist do?

Interview with Christian Donnerer, Data Scientist at Faculty

What’s your job title, and what do you do?

My job title is Data Scientist. I do Data Science all day long.

What does your day to day look like?

Today, for example, wasn’t particularly interesting. Usually we have a stand-up with the project team in the morning. However, on Friday mornings you are given time for self-directed learning, and to try things that you are interested in. I didn't have time for that today, as I was working on generating training material for a client about Machine Learning Ops. I came up with an example of how to serve a Machine Learning model in a reliable and robust way. Then I had some project meetings. That’s largely it - maybe two hours of meetings, and the rest of the day was spent writing code and answering some ad hoc requests from colleagues. It wasn’t a particularly exciting day.

What percentage of your time do you usually spend coding vs meetings and other work?

It's not a lot of coding! Maybe 20 % on average programming, 40% of meetings and the remainder anything from Slack / email to making docs and slides. If I can get four hours of actual productive coding in a day, I'm very happy. That doesn't happen very often, typically it's an hour or two. That doesn't mean I don't work for the rest of the day - it’s just that meetings, planning and researching about the problem tend to dominate my day. In four hours of coding you can get a lot done, if you have identified the right problem to focus on, which often is the hardest part.

How did you get into data science?

I did a PhD in experimental Physics at UCL. While I quite liked Physics, I realised in my fourth year that I don't want to do it permanently, because it seemed like a tricky career choice that doesn’t provide job security - essentially you’re somewhat reliant on luck that your experiments produce interesting research!

Hence I applied for a few positions at the end of my PhD, including a Postdoc and a 6 week Data Science “conversion” course. I got accepted to the latter and thought “Let’s try it!”, mainly because it seemed like something different and new. If I really didn’t like it, I could still do the Postdoc. So I did the course, and quickly started to realise that I wanted to continue with Data Science, as opposed to academia. One of the things that stuck out was that I can go anywhere in the world and just do this job as a Data Scientist, I don't have to go to some strange research centre in the middle of nowhere!

However, six weeks is definitely not enough time to become a Data Scientist, so I self studied for another three or four months, while also finishing the corrections to my PhD. At that point I was in a position to be reasonably confident in applying for Data Science jobs. I failed quite a few interviews - perhaps I applied to around 10 and had a couple of offers at the end - so it wasn’t super successful. One of the offers was at IQVIA, I liked the team there, and soon afterward started working as a Data Scientist. I worked there for about 2 years, and then moved to Faculty, where I am now, again almost two years ago.

What do you find to be the most meaningful part of your job?

In terms of finding meaning in projects, I feel like the best projects are the ones that are a little bit outside of the usual. I worked on a really interesting project for a railway company, where we used computer vision to automatically detect whether a tree or other vegetation is encroaching on the railway tracks. That could make a difference in terms of what people do on the ground - essentially prioritising where to cut down trees. For me, “meaningful” is when you achieve impact, or help automate tasks that humans don't really like to do. That isn’t always the case in my work, sometimes the projects are along the lines of who to show ads to so we can get people to buy more things, which isn’t really that meaningful.

I'm most interested in projects that are “0 to 60” - where nothing currently exists and you create a solution (going from 0% to 60%), as opposed to taking something that’s already running at 80% and making it a little bit better, getting it to 81%. That "0 to 60" type of problem is something you can often find at Faculty, where we get to work on genuinely new problems that people haven’t attempted before.

Overall, I don't really feel the need to do work I want to be proud of. There are a lot of interesting problems out there, I feel like doing those well is enough.

What are you most proud of in your Data Science career?

I enjoy doing things well, like writing high quality code and making sure it keeps running in the future. Good engineering work is probably something I'm proud of - how do we design code and systems that are reliable and robust? This becomes especially difficult if you work on consultancy type projects with short contracts.

Overall, I don't really feel the need to do work I want to be proud of. There are a lot of interesting problems out there, I feel like doing those well is enough.

Do you ever suffer from an imposter syndrome?

Oh, that's a good question! I used to, a lot, when I was in academia.

I’d always think “Oh, everyone else knows more about this than I do”. I also used to think that early on in my Data Science career, but not so much nowadays.

I think I have reached the point where it's becoming clear that most people, most of the time, don't really know what they're doing. Sometimes they can be very confident and look like they know what they're doing. And then other people look up to them because they seem confident.

For example, let’s say you know how to deploy a Machine Learning model in for a specific project. Suddenly, you are known as the expert on deploying Machine Learning models - but what if you need to use a different technology for a different use case? I feel in my sector, there are a lot of people who are experts on specific techniques , but quite often, they don’t really know enough about the wider domain - there are always so many ways to approach a problem! After realising that, I don’t really feel impostor syndrome as a Data Scientist that much.

Although I’ve felt it a bit recently because I’ve started becoming more of a tech lead. Maybe Impostor Syndrome is something that happens when you’re new in a role and don't quite know what's expected of you.

What advice would you give to someone who wants to become a Data Scientist?

A lot of people will come into Data Science from a similar background - Physics, Maths, Biology, Chemistry, Statistic, Economics… But that doesn't mean that if you don't have that background you shouldn't do Data Science!

If at some point you are curious “What’s this Data Science thing?”, the best thing you can do is get into one of those Data Science training / conversion courses where you get to work on a real project. While it’s not necessarily the best way to learn, it is the fastest way to build up your network, which is ultimately where most job offers come from.

There are plenty of great courses on Coursera, for example, you can do an entire Data Science specialisation - most people will be perfectly capable of mastering these topics. In 6 months to a year, depending on whether you have prior programming experience or not, you can probably pick up enough Data Science skills for a junior Data Scientist role.

The tricky thing is then getting your first position. The conversion courses really help there, because they give you practical experience, and the networking side is just so valuable. You might be hired for a position you aren’t 100% qualified for (yet!) because you have some connections. That’s okay - most people learn substantially on the job when they start out.

If you aren’t able to do a conversion course, focus on good side projects. Don’t do the standard ones that classify cats and dogs. Instead, find something useful and new, and really think it through end to end, from the raw data to a deployed model. Go deep on one or two projects like that, that can show off your skills - it will make a lot of difference in interviews.

Doing some practical projects will also help you get a better idea of what Data Science is - you won’t quite know what it is until you’ve done it!

bottom of page