Julie Gould:
Hello, I’m Julie Gould and this is Working Scientist, a Nature Careers podcast. This is the third part of our series on careers in physics, where we’re exploring transitions. Last week we heard from Elizabeth Tasker, a UK-born astrophysicist who transitioned to Japan and now combines her love of research with her love of science communication. But in this episode, we’re exploring a slightly different type of transition – from physics to data science. It’s a topic that I’ve been keen to explore because physicists are often coveted by industry for their skills in data science, but there are so many more people graduating from data science-focused graduate degrees that I wonder if there’s still a place for physicists in this industry. So, this is exactly why Kim Nilsson set up Pivigo in 2013 – initially, to help academics transition from academia to industry, but now with a special focus on data science. I spoke to Kim to find out more about her background in astrophysics and why and how she transitioned to business, and the conversation started when I asked her why she decided to leave academia in the first place.
Kim Nilsson:
It was during my PhD studies when I first started to think hang on, I’m not sure this is the right thing for me because I realised that the things I enjoyed doing the most were actually the project management elements and the making things happen element, rather than the actual technical elements of sitting in front of a computer and coding all day. And so, it started to put a seed of doubt in my mind during my PhD whether it was going to be the right career for me, but I then pursued another two postdoc positions after that because it was a lifelong dream and it’s not easy to let go of that.
Julie Gould:
I can totally agree with you there. It is really hard. I know many PhD students and postdocs who have exactly that feeling, that they’ve worked so hard and they’ve always wanted to do this and even though they know that they’re potential prospects in academia are limited, they don’t want to leave, and then there’s also the fear of being looked at as a failure when considering options outside of academia. Was that something that ever crossed your mind?
Kim Nilsson:
The failure bit – not so much. I think it was more just a fear of the unknown, of taking that jump out of academia, which was the only thing I had ever known, and having really no idea of where I was going to land. I just had to trust myself that I would figure it out somewhere along the way.
Julie Gould:
So, you made the jump. You left academia after your second postdoc. So, where did you go and what did you do?
Kim Nilsson:
I spent about a year applying for just other jobs and jobs within project management, both within science and outside of science, and for other consultancy jobs, management consultancy, strategy consultancy, but I was completely unsuccessful in all of those applications, which of course, really threw me because again, you start to doubt can I really do this. But after that year, I was really bitten by the business bug and I really thought my future is somewhere within business but I’m not quite sure where and therefore I decided to do the MBA and I figured if I have a PhD and an MBA surely someone will want to hire me.
Julie Gould:
So, do you have any thoughts about why you were so unsuccessful for that year where you were looking for jobs?
Kim Nilsson:
I think this is very related to also what we see in the PhDs that we hope to transition into data science today. It’s that when you have spent your life in academia, you are totally unprepared for what business life is like and I mean in terms of communication, in terms of teamwork, in terms of the softer skills, and you have this very academic mindset which many of these companies just do not appreciate, unfortunately, and so it requires a change in mindset and there are many ways to do that but I think I was just too academic in how I came across in those interviews.
Julie Gould:
So, you went on, you did an MBA – did you enjoy it?
Kim Nilsson:
Absolutely, yes, it was a fantastic year.
Julie Gould:
What often happens to people who do things like an MBA degree is they have a seed of an idea of a business they might like to set up. Is that something that happened to you?
Kim Nilsson:
And then about halfway through the programme I met Jason, who is the cofounder of my business, who had a recruitment background before the MBA and together we started to think about all the challenges that me and my friends has faced in making this transition. We started thinking about an industry that is constantly saying that they can’t find enough analytical talent, and we thought there was a gap to be bridged, where we could be really passionate about supporting academics in making that transition.
Julie Gould:
Transition into…
Kim Nilsson:
Initially, it was anything really. We just wanted to help PhDs get jobs. But very quickly after that, we then zoomed in on data science. This was now about seven years ago. It was a new thing. It was just around the time when Harvard said that it was the sexiest job of the twenty-first century, and lots of job opportunities and something that also not very many academics knew about at the time, and so it was an area that we got excited to work in.
Julie Gould:
It’s funny that they didn’t know about it because there are so many scientists that pretty much all of what they do is data science, especially in a subject like astrophysics.
Kim Nilsson:
Very interestingly, in those first couple of years when I would go out to universities and give talks and presentations on careers outside academia, I would ask them what roles they were aware of that they could do and it tended to be finance, it tended to be software development, IP etc., but when I said well, have you heard about data science and this is what the jobs would be like and this is the salary you would get etc., they couldn’t believe their eyes. They were shocked and very, very pleasantly surprised that this option existed, and then they all got very excited about it.
Julie Gould:
So, what makes a scientist so suitable for working in data science in industry?
Kim Nilsson:
I think, especially when you come from a physics background, you will already have done a lot of coding, a lot of software development, so you already have those skills. Secondly, you will already typically have worked with large datasets, with analyses, with maths, statistics, and those are the two largest groupings of skills that you need to be a data scientist. And on top of that, what you then have is this scientific mindset which actually is important in data science because in data science you need to have a hypothesis, you need to set up an experiment, you need to run it, you need to then be able to critically evaluate the results that come out, and all of these are scientific skills. So, in principle, physicists are the full package.
Julie Gould:
So, what sort of training do you run at Pivigo for scientists who want to become data scientists?
Kim Nilsson:
About six years ago now, we started these Science to Data Science programmes (S2DS) and the whole idea was that okay, PhDs, they have these amazing skills already. What they’re lacking is that little bit of commercial polish, as I mentioned, the understanding of how to use these skills within a commercial environment. And so, we built this programme around well, let’s bring together these super smart, super motivated PhDs with companies who want to hire and are interested in data science and get them just to work on a project. So, for five weeks during S2DS, our participants work on a project with a company. They deliver the project as if they were consultants, and they get that experience in a very safe and risk-free environment and it will help them then go out and apply for a job full-time after that.
Julie Gould:
Now, one of the aspects of your Data to Data Science training programme is that you do some video conferencing so people can do their training programmes from home. Do you find that there are women, particularly with young children, for example, that take part in this because they have young children but they really want to make this transition from science to data science?
Kim Nilsson:
Yes, initially our programme was only based in London physically onsite on a campus, but we then decided to start a remote virtual version of the programme, and one of the key reasons for that was because we know there are some people who just can’t travel, who can’t spend five weeks away from home, and so what we see is often the people who do the remote version indeed do have other responsibilities, typically parenting responsibilities. I have a great story, once, about how we were on one of these video calls with a team, discussing the project and it was a very professional conversation. One of the women sat a little bit awkward, but I didn’t think much of it until her husband came up from behind her and picked up the baby that she had been nursing while having this conversation and it blew me away how we are providing an opportunity here for someone who otherwise would not be able to do this, and it was a proud moment both for me and for her.
Julie Gould:
It sounds like something that I enjoy seeing as well. I’ve been on many panels at conferences and more and more you see women bringing their young children to these conferences, and I’ve even at a few occasions seen women bring their baby up on stage and they’ve had to nurse during a talk, so it’s fantastic that you’re able to offer this opportunity as well. Thank you very much, Kim. So, you’ve actually bought your colleague with you, Deepak. Deepak, can you quickly introduce yourself?
Deepak Mahtani:
My name is Deepak Mahtani. I’m the community manager and data scientist at Pivigo.
Julie Gould:
So, you’ve actually been through the programme that Kim set up.
Deepak Mahtani:
Yeah, so I was actually on the virtual programme in March 2016.
Julie Gould:
And why virtual?
Deepak Mahtani:
It was the one that came about when I was free. I finished my PhD in January and the next available programme was March, so it was just the right timing.
Julie Gould:
And why did you decide to transition after your PhD?
Deepak Mahtani:
Well, towards the end of my PhD, I was thinking about applying for postdocs and so forth, and I applied for one or two, but then the more I spoke to colleagues who were already there, it became very apparent to me that I’d have to move around every three of four years and also, I might have to move country. I might have to move half way across the world and I wasn’t prepared to do that yet. I had a very elderly grandmother at the time and I wanted to start settling down. So, a friend of mine told me about data science and the S2DS boot camp, and the more I looked into it the more fascinated I became and realised that data science really takes all of the bits I loved about my PhD without all of the stuff I didn’t like.
Julie Gould:
So, what did you do in your PhD?
Deepak Mahtani:
So, I studied exoplanets, so these are planets around other stars, and specifically I was looking at their atmospheres to try and understand how they work and the chemical and physical properties of them.
Julie Gould:
And that requires a lot of data processing?
Deepak Mahtani
Yes, so I was very fortunate to use a space-based telescope called Spitzer which gets hundreds of gigabytes of data, and there was just loads sitting in the archive that I was able to analyse, specifically two specific stars, and the time it takes to analyse the data, it’s on the timescale of months. But there’s a lot of it and you gain a lot of really interesting skills from just simple coding to asking the right questions of your data to really critically analysing the results that come out, and those are the key skills that you need for any role specifically within data science.
Julie Gould:
So, tell me a little bit about the Science to Data Science programme and what is was like for you going through that programme.
Deepak Mahtani:
Sure, so I came into it with having just about picked up Python and was terrified because I had learnt a very under-utilised language outside of academia, and so this was this brand new programming language and then I was told to build this fancy recommendation engine and I was like oh my god, what do I do now? But one of the best things about the programme is that you’re in teams of three of four people and so you’re able to utilise your strengths and understand where your weaknesses are. And from there, it became really apparent to me that actually through just a bit of googling and trial and error, you can get to where you need to. And just like Kim was mentioning earlier, changing your mindset from that perfectionist mindset of it has to be right first time, to just get it working in a hack and slash way for now, and then once that’s done you can tidy that up and make it faster and more efficient, and that was how we did it. It was good fun.
Julie Gould:
So, when you completed the five-week course, what happened next?
Deepak Mahtani:
So, I think we actually finished it on my birthday.
Julie Gould:
That’s a nice way to finish.
Deepak Mahtani:
It was, it was really good fun, and then I went on to work at a gambling company and I didn’t really enjoy it very much so I left after about seven months, and about a week before, I’d spoken to Kim, and I was like, ‘Kim, I’m not enjoying myself here.’ So, she said, ‘Come into the office’, so I was like okay. I trundled into the office after work one day and she said, ‘Well, the community manager role has become available and you could do some data science there too.’ And I thought okay, so I went home and I thought about it and I read into the job spec and realised that it combines both my love of technical stuff but also my communication and people skills, and so it was just the perfect job at the right time. So, I applied for it, interviewed for it a week later and had the job on the Friday, so it was a very interesting experience and now I get to travel and talk to loads of PhD students to give them more of the advice that I wish I’d got. I mean I was very lucky that I had someone to talk to and get advice from when I was making my transition, but not everyone has, so I get to go and speak to everyone and give them all of that advice and help them make that transition really smoothly.
Julie Gould:
So, what sort of advice would you offer to those who are looking to transition?
Deepak Mahtani:
Well, I try to give tangible advice. So, a lot of times when you look online, it’s just sort of generic do this and do that, but I try to tell them that there’s three main things you should do. You should learn about either Python or R because they’re the two most used programming languages within data science. Read up a little bit on machine learning in that you don’t have to know about every algorithm under the Sun, but understand the differences between, for example, supervised and unsupervised learning and what the difference between classification and regression are. And then understand a little bit of SQL as well because a lot of data is stored in some kind of database and so you really need to be able to access that data and the simplest way to do that is through a relational database which uses SQL. And I also recommend two books that really helped me to understand how to change that mindset from academic to business, which were Crucial Conversations and the other one was called Just Listen, and those two books, what they really do is show you how to be empathetic and understand what your stakeholder is looking for, why they need it and when they need it, and also understanding how to manage those expectations. It’s really important that a lot of stakeholders in the business world might want something tomorrow and you can try and deliver it maybe not tomorrow but the next day, but manage those expectations and those two books really helped me.
Julie Gould:
Deepak, thank you very much.
Deepak Mahtani:
You’re welcome.
Julie Gould:
Now, someone else who’s made the transition from physics to data science is an old university colleague of mine, Lewis Armitage. He completed an undergraduate masters in physics at Cardiff University with me before moving on to do a PhD at Queen Mary University in London. His work was partly based at CERN, the European organisation for nuclear research in Switzerland. Now, he decided, like Deepak, that he needed some more work-life balance and also thought that data science would be where he’d find that. Here’s his story. When you were working on your PhD, you had the opportunity to go out and work at CERN. That must have been super exciting to then be able to go to basically the home of particle physics.
Lewis Armitage:
Yeah, exactly. I mean it was actually so amazing that I didn’t quite believe it myself and I think that actually, my family didn’t really think that I’d ever be able to get there. I can actually remember telling my family that I was actually going to apply for this PhD and I was hoping to get it and they were like, ‘But Lewis, no one works at these institutions. That’s crazy. Only crazily good people work at these institutions.’ And I was like oh, thanks guys, thanks for your confidence. Laughs. But it turns out that physicists from everywhere can work at these institutions because we’ve got really, really good skillsets.
Julie Gould:
When you got to CERN, what was it like to actually work there?
Lewis Armitage:
Well, I was actually quite surprised really because it is an extremely large organisation and then I was working on the ATLAS experiment, which has hundreds and hundreds of people working on it, and you never really meet everyone who works on the experiment. It would be almost impossible to meet everyone.
Julie Gould:
I find it interesting to think that you’re part of a team where you never actually meet everybody on the team. Did it make you feel like, even though you felt like a superstar having been given the chance to work at a place like CERN, did it make you feel very small?
Lewis Armitage:
Yeah, I think it does and I think when you start off, that’s always going to be the impression that you get because everyone there knows a lot of other people there. It’s kind of like your first day of school, you know, you’re there and you’ve got to meet everyone else, you’ve got to make your network. And then everything seems very big in the way that other people already have their analyses to take care of, their own responsibilities, and you’re still kind of finding your feet. But then actually, as the weeks go by, you get more confident and your analysis gets a direction and then you start plugging into these different teams to actually start getting information that you need to move forward. And then towards the end of the PhD, you feel like actually, you know what, I’ve got a place here.
Julie Gould:
Feeling like the superstar you felt at the beginning when you were accepted. So, what happened next? You decided to make a move into industry.
Lewis Armitage:
Yeah, there’s quite a few things that happen in a large organisation such as CERN. One of the, perhaps, downsides is that because there’s a lot of people who work there and there’s a lot of people who are trying to make their name in science, there becomes an element of competition, I think, and it really pushes people to work as hard as they can, and I think that’s really, really good. But it’s got this downside in that you start to give your whole life to the subject. This was something I was noticing really, in that it can be difficult to switch off from the work that you do, from the physics that you’re trying to do, and so, you’ll notice that all of your evenings become occupied and that becomes routine, and then beyond that, all of your weekends are becoming occupied and that’s routine, and I saw this as actually a really unhealthy work-life balance.
Julie Gould:
So, it wasn’t a lack of love for the subject.
Lewis Armitage:
Even though I really enjoyed what I was doing, I couldn’t bring myself to do it every day and to not switch off from it. I really wanted to have my own weekends for myself. I wanted to get back home and just speak to my friends and talk about something completely different.
Julie Gould:
So, after CERN, you moved to industry. You chose a path of data science. Now, what was the job hunt like?
Lewis Armitage:
Yeah, it was quite difficult because I don’t think I really appreciated what you need to do for look for a job. I mean it sounds kind of simple. It comes down to the really basic things like how do you write a CV, how do you write a cover letter, what kind of jobs are interesting, where you should kind of target and position yourself, even how to read a job description is actually really important, and although I had these really strong skills, it was difficult for me to market them properly because I didn’t really know what the businesses were actually looking for and what was actually actionable from my skills, and so that was the thing that I learnt very slowly, actually.
Julie Gould:
Why do you say it was a slow learning process?
Lewis Armitage:
I think being naïve, I think I sent out a load of applications and then I just kind of sat back and thought okay, that’s it, I’ve sent out all these applications, that’s done. And then it’s only when you kind of start only getting a few replies and then they don’t really go anywhere that you actually question yourself and go actually, maybe I’m not as strong as I think I am and then maybe I’ve actually got to review myself and then you modify your CV and your cover letters and the style of it and then you send them out again, and then you get a bit better responses but then it still mostly comes back negative and then you think what it is about this and you can turn to your friends and they can make suggestions for you.
Julie Gould:
How long did it take you to find a job?
Lewis Armitage:
A little less than a year, actually.
Julie Gould:
What advice should people be following who are interested in a position that is heavy in data science and is in industry?
Lewis Armitage:
If you’ve shown that you’ve actually been able to take data and produce results from your data and then interpret that data – and the key thing is interpret – then that would really be the thing that puts you above because physicists have very good critical thinking skills. But then being able to justify that for a data science position, it really depends on the position. It depends if the data science position is actually a half analyst position. If that’s the case then the critical thinking will come in immensely, but if it’s just a data science position that’s more like full stack developer or something like this where the candidate is meant to do the data warehousing, they’re meant to create all these APIs and then also do some data cleaning and data manipulation for some end user or some end result, and it’s really the end user who then looks at the data and decides whether it makes sense or not and then they will feed that information back to the data scientist, If that’s the case then physicists are at a disadvantage there, and that’s really not, in my personal opinion, that’s not the place where physicists should be going because it’s unlikely that you’ve got the data warehousing skills. It’s unlikely you’ve got experience building APIs. I mean maybe you do and that’s good. And so, I think this is a key thing with, again, reading the job description.
Julie Gould:
So, you are an analyst at Tsquared Insights in Geneva. So, what does an analyst do?
Lewis Armitage:
So, for my day-to-day job, essentially, I take data that’s already been processed by an RND team who are full stack developers, and then I have a brief that is the client’s requirements and I’ve got to satisfy those requirements for their analysis and I’ve got to build an analysis around the data. So, I’ve then got to write the code which will then access the database, it will then process the data in a particular way. It will chop up the data into the right components and then it will run various statistical analyses, again depending on what the client wants, and then I’ll output a certain number of files. I take those files and then I put them into some deliverable, whether it’s a presentation or some Excel file perhaps that the client wants. But then there’s a key element there at the very end which is to look at your results and look at the data and to make some insights about it. You’ve got to look at the data and go okay, what’s actionable here? What will the clients find useful? What is going to make us as a business look really good with our data? And then that’s really where I inject my creativity and I inject the critical thinking because that is something that not everyone can do.
Julie Gould:
Thanks to Lewis Armitage. Now, in the next episode, I speak to Professor Jon Butterworth from University College London and he works at CERN just like Lewis did. He spent many years working on the ATLAS project and supervising students who have done the same. Now, I wanted to speak to him to find out what it’s really like working on such an enormous, international team like ATLAS, which led the discovery of the Higgs boson, especially when there was such a huge media focus around it. Here’s a sneak preview.
Jon Butterworth
One of the nice things with particle physics is it’s not all down to one PI and their lab. There’s a huge number of us, so it was good that wherever anyone in the media pointed their microphone, they found someone who was excited because the excitement was real. But it was also good that, well, some physicists’ worst nightmare is to be in front of the camera and that’s absolutely fair enough. Everyone doesn’t need to do it.
Julie Gould:
Now, don’t forget you can always find out more about what the Nature Careers team is up to on Facebook and Twitter, and there’s of course the website – www.nature.com/careers. Thanks for listening. I’m Julie Gould.
