IKEA DATA SCIENCE LEADER TO FUTURE Ph.D’s: CONSIDER INDUSTRY POSSIBILITES INSTEAD
Emil S. Jørgensen is Director of Product Analytics at IKEA, leading the development of data-driven digital products in the Customer Experience Domain. This includes, among other things, all machine learning features on IKEA.com and in the IKEA App, as well as their increasing ability to personalize campaigns across digital marketing channels. He joined IKEA a little less than 2 years ago as the second employee in a newly established Data & Analytics function; a function that today consists of 100+ leaders and individual contributors and is still growing at rapid speed!
Before joining IKEA, Emil finished a Ph.D. in probability theory from the University of Copenhagen. This makes him a great candidate for assessing the value of pursuing a Ph.D. within the realm of applied mathematics and machine learning. In this post, we dive into the motives Emil had for pursuing a Ph.D., how life as a Ph.D. candidate is in reality, and why it makes sense for someone with a theoretical Ph.D. to play a central role in the digital and data transformation that IKEA is currently going through.
Finally, Emil gives his view on the aspects you should consider, to decide whether a Ph.D. path is the best choice for you – and why he believes a Ph.D. in applied mathematics and machine learning may be something of the past.
This post is filled with great advice, especially if you are considering doing a math-centric Ph.D.!
Career advice is always subjective and these choices are, at their core, personal preferences. However, there is still tremendous value to be found in mimicking and learning from interesting people with interesting careers – especially those whose story you can see yourself in.
Conflux Insights is a new series of articles where we learn key insights and findings from interesting Conflux mentors from the Danish STEM industries and share them with you.
Emil graduated as a MSc in Mathematical Modelling and Computation from DTU in 2014. As a newly graduated engineer, he considered multiple options: data science, management consulting, and quantitative finance, to name a few. However, having finished his MSc at DTU, he still felt a large gap in his mathematical understanding of more applied topics. Too much of the theoretical foundation had been skipped, he felt. This led him to pursue a Ph.D. in probability theory at the University of Copenhagen.
After his Ph.D., he joined KAYAK, one of the world’s largest meta-search engines in the travel space. Here he worked as a Data Scientist for roughly a year, before joining IKEA’s newly created Data & Analytics function.
Why Emil Chose a Ph.D
After 5 years of applied mathematics at university level, Emil still sat with the feeling that there was a greater picture he was probably missing. Axioms and fundamentals had been glossed over, leaving him with a gap in his knowledge. “It wasn’t an easy choice to do a Ph.D. I love the engineering mentality you are taught at DTU, but after studying abroad at ETH Zurich for a year, I realized that I had a huge theoretical gap in my understanding of mathematics. And it frustrated me.” he explains.
In my mind, a Ph.D. in theoretical mathematics was something reserved for the truly hardcore math wizards – the type whose dreams may even be mathematical proofs. And if you were to only look at Emil’s very math-heavy résumé, you might think that he would fit that pre-judgment; this, however, turns out to be far from the case:
“Actually, I’ve never seen myself as particularly talented, especially when I compared myself to my peers at the time. I knew I was a good and hardworking student, but what blew my mind was when I went abroad to ETH Zurich and saw how some pure mathematics students could reason and playfully carry out proofs. By studying at ETH, I went from comparing myself to my Danish peers, to compare myself with the entire world. And I believe that is a very healthy experience to go through during your studies. I highly encourage everyone to go abroad – and especially to seek towards the very ambitious universities. This will change your mindset and allow you to think bigger.”
Before talking to Emil, I presumed that he had mainly chosen to pursue a theoretical Ph.D. because he valued the beauty of mathematics in and of itself:
“Yes, I absolutely value the beauty of complex things by themselves. But from day one of my Ph.D., I knew that I didn’t do it to enter the world of academia. It was always about becoming better at solving problems. And I was sure that I would become better at any form of quantitative reasoning by pursuing a Ph.D. I wanted to take my decision-making abilities up a notch by truly understanding the theory of my domain.” When looking back now, Emil believes that he achieved this.
The Ph.D. Life
Some may see a Ph.D. as the next natural step when you graduate from your master’s degree, so I was curious to dive into how Emil’s Ph.D. work life looked like. Was it just an extension of the typical work-life at university?
“In my opinion, you should not view the Ph.D. program as a natural extension of your university degree. In Denmark, the Ph.D. is typically shorter than abroad, which makes it quite intense. It is expected of you to produce scientific results in only 3 years, and you are treated as an independent and responsible researcher, instead of just a student. This premise is a major step up from what is expected of you during e.g. your master’s degree.”
The first year was all about understanding “the basics”, i.e. figuring out where a potential research contribution could be made. This meant that most of his time was spent reading articles in state-of-the-art journals and studying results and proofs to understand the domain in general. “This was a quite frustrating and difficult solo-experience”, he explains. The purpose of this (dreadful) period is to get you up to speed with recent theoretical developments so that you become comfortable enough to start brainstorming on and navigating your own ideas. “After many months of intense hard work, mathematical theory can start to feel natural to you, and you will be able to move more freely”, he explains. And it is completely essential for the success of your Ph.D. to reach this point because, at this point, you realize that great research is about so much more than just studying theory:
“Great research is an immensely creative process, and this becomes clear to you when you start to understand your domain to its core. Great researchers are truly creative – even though (or precisely because) they work under deep technical and theoretical constraints. This was one of the main insights I obtained during my Ph.D.”, he says.
But his own story turned out to be a bit bumpier than originally anticipated. As part of the Ph.D. research program, Emil visited the University of Chicago to continue his work around 1,5 years into the program. Here he was placed in an empty office all by himself. “And when you’ve got that much alone time on your hands, you start to think things through and, in my case, connect the many pieces in front of me. I realized that the current topic I was working on, would not lead anywhere, and so I made a drastic decision to pivot!”.
Pivoting this late is a terrifying experience, and the fear of admitting to yourself that all your work should be thrown away is dreadful. But to keep going down the current path is to dig the hole deeper. This leads to the question: how do you even know for sure if pivoting is the right move? If you find yourself in this position, you should check out the phenomena book The Dip: A Little Book That Teaches You When To Quit (and When to Stick) by Seth Godin. Worth a read!
The pivot turned out to be a great decision, as Emil during the last 1,5 years of his research program managed to find the right balance between his own interests, and those of his supervisor. Together, they started to dive into unknown territory related to estimating functions, a topic that his supervisor has been pioneering for more than 20 years, and a place where they could bridge general asymptotic theory with Emil’s fascination with high-frequency financial econometrics and statistics. This led to 2 (lengthy!) consecutive research papers, one of which has just been published on arXiv and submitted to a scientific journal for review. The 2nd contribution is planned for publication soon. Versions of both papers can also be found in the final Ph.D. thesis here.
Is a Ph.D. in Machine Learning Still Viable Today?
According to Emil, we’re witnessing some very interesting changes in the relationship between academia and industry when it comes to data science and machine learning specifically. For the past 20 years, these areas have been the study of massive research communities – largely stemming from corporations, instead of purely academic research groups. There are multiple examples of ground-breaking research teams from industry, be it Open AI, Google DeepMind, Alpha Go, or IBM Watson, and much of the Data Science development toolkit is also driven as open-source frameworks by private institutions. Strong examples of this are Google TensorFlow, and Facebook’s PyTorch.
“The data mining techniques that advance within data science are built upon, originate from academia. However, with progress in modern machine learning being powered primarily by access to vast amounts of training data and computing power, tech giants are finding themselves in possession of a huge competitive advantage – especially because the same giants are also non-incidentally the ones sitting on massive cloud computing infrastructure. I think this is the main reason why the industry is arguably taking a comfortable lead from academia.”
Previously, you often had to enroll in a Ph.D. program if you wanted to dive deep into a technical topic. But now that some industries have realized the business- and people potential of running large, mature Research & Development (R&D) departments, you may find the same, if not better, opportunities by joining them instead. And the universities are also starting to notice this shift, according to Emil:
“Today, you typically see a more connected, intertwined relation between academia and industry, in which projects are driven as a hands-on collaboration between both parties. And it’s becoming more common to see top researchers working part-time in industry, alongside their university careers – if not joining an industry position full-time.”
Of course, a Ph.D. degree in Applied Mathematics is still a great career investment to consider. But you would fool yourself if you were to not also investigate the possibilities of achieving your goals by joining a large, research-heavy technology company instead.
How To Figure Out If a Ph.D. Is Right For You
Enrolling in a Ph.D. program is a big decision to make, and the experience will be quite different from anything else, you’ve probably tried previously during your studies. 3 years of your life depend on this decision, so it is crucial to make the right call. So how do you figure out if it is the right move for you?
By running a “Mini-Ph.D” experiment! Many universities worldwide run programs that allow you to join a research group for a few months to try it out – just like the program Emil did in Chicago. This allows you to get a small appetizer for the real deal without having to spend more than a summer abroad – which, in itself, is a tremendous experience!
Sune Grønskov, General Manager at Ørsted, has also made heavy use of this advice whenever he was about to transition into a new role; something you can read more about here.
But be aware that this advice is prone to a phenomenon called the hindsight bias, meaning that after the experiment you will probably believe the outcome to have been obvious all along – even though it wasn’t!
Let me explain; imagine that you consider doing a Ph.D., but are unsure if it is the right decision for you. To find out, you join a research program during the Summer break. From the program, you either find out that a Ph.D. is the right or wrong fit for you – and in both cases, you will then take the right decision moving forward. But paradoxically, both cases will most likely leave you with the feeling that you knew this outcome all along, and that joining the program was a waste. The productivity guru Scott Young explains this phenomenon nicely here.
The same advice also holds if you consider joining the R&D department of a prominent technology company. Many of these firms offer internships that allow you to try out the experience before committing to it.
When Emil looks back at his time during the Ph.D., there are some character traits which in particular improved the experience. Most noticeably that he had a lot of self-drive and discipline. The course of the Ph.D. required him to do difficult work on his own for extended periods, and it was key for Emil to be self-managed:
“I’ve always been driven by the outcome of hard work. It fulfills me when I finally understand a difficult theory, or finish a mathematical proof that I’ve been working on for days”, he says. I know that many of us in the engineering world own the same trait; to learn something technical, is satisfying on its own.
This is the first of two posts from the conversation with Emil. In the second one we will dive into Emil’s experiences in the Data Science world – both as in Individual Contributor at Kayak and as a Leader at Ikea.
This is the third of several articles from Conflux Insights. Our goal is to learn from real career stories, extract key findings from them, and share them with you.
If you have any comments or questions about the article or Conflux Insights in general, please reach out at email@example.com.
Thanks to Emil S. Jørgensen for sharing his experiences and career history with us.
This interview was performed and written by Jakob from Conflux.