
Aras Selvi

“Needless to say, London is the heart of the business world, and Imperial College London has a great business network. What makes Imperial special for me is the science and engineering focus. I have a background in engineering and applied mathematics, and I am very happy to see that here at Imperial. ”
Read Aras's story
Why did you decide to study the Doctoral programme at Imperial College Business School and what makes this programme unique?
There are many reasons why I decided to study at Imperial College Business School. Firstly, the London factor. As a Business School student, it is very important for me to be engaged with businesses during my PhD years. Needless to say, London is the heart of the business world, and Imperial College London has a great business network. Secondly, what makes Imperial special for me is the science and engineering focus. I have a background in engineering and applied mathematics, and I am very happy to see that here at Imperial, we have an interdisciplinary programme. I am in the Business School, doing research in optimisation and theoretical computer science which is important for businesses, and taking extra modules from the Electrical and Electronical Engineering department!
Finally, the PhD Supervisor. I was reading the papers of my current supervisor, Professor Wolfram Wiesemann, before applying to Imperial. He is a Professor of Analytics and Operations at the Business School. I was really impressed by his work in the field of ‘Distributionally Robust Optimisation’ and I wanted to be a part of his research group.
What is your previous academic and work experience and how did this prepare you for the programme?
I obtained a Bachelor of Science degree in Industrial Engineering at Ozyegin University (Istanbul, Turkey). Here I also did a minor program in Computer Science, which helped me to write codes of the analytics and optimization projects of mine. I did two internships at Coca-Cola Icecek. Thanks to my Bachelor studies, I was able to develop an optimisation algorithm for the sugar procurement strategy of the company, and to write a software which takes input from the users and output the decisions. After seeing the pleasure of taking knowledge into practice, I wanted to improve my knowledge in the theoretical side of my field (namely mathematics of operations research). I went to Tilburg University (the Netherlands) where I did a research Master’s programme in Operations Research and I was a member of the Dutch Operations Research Network. The programme was designed to prepare the students for a PhD in this field and I had the chance to be taught by great professors. My research project was in ‘Adjustable Robust Optimisation’. During this study, I was reading the work of Professor Wolfram Wiesemann, hence I wanted to continue my studies in Professor Wiesemann’s research group. I graduated from Tilburg University with a Cum Laude degree (equivalent to 1:1 honours) and came to London.
"Needless to say, London is the heart of the business world, and Imperial College London has a great business network."
What has been your favourite module?
My favourite is module is Data Analysis Tools by Dr Harjoat Bhamra. This is a module that all the MRes students take at the beginning of their programme, which gives them a mathematics and statistics background (or refreshes it) at a fast pace. Almost all of us knew about these topics already, but most of us forgot these since we took some topics in the first year of our undergraduate studies. The way Dr Bhamra teaches this module has made me never forget these materials - he has a Maths background and he really has a great intuition of the materials.
What area of research will you be doing your PhD on?
Very broadly, optimisation, theoretical computer science, and machine learning. My main plan is to bring an optimisation perspective to problems in theoretical computer science and machine learning. For example, in machine learning, a lot of methods apply optimisation. The famous PCA problem optimises the variance explained by PCs, cancer detection algorithms use SVMs which optimize the classification given by a designed Kernel, Neural Networks optimize the parameters of the network, and so on. However, these are automatized in many packages of software today, and one needs to know the optimisation algorithm lying behind to ‘manipulate’ the model for different purposes. For example, what if the data we train our algorithm has measurement errors? Are we going to optimise the false data? Or are we going to develop an optimisation model which ‘knows’ that some of the data is perturbed. What if we are optimising an algorithm which has data coming from surveys, and the users potentially lie? These are very well-known phenomena in the machine learning field, but my research extends these questions.
Another research topic of ours is interpretable machine learning (which is a very active study these days). Some methods like neural networks are black-box models, namely they work well for detecting & understanding things, but not good at explaining why. If we can derive the equivalent mathematical models lying behind these algorithms, then we can explain maths!
"What makes Imperial special for me is the science and engineering focus. I have a background in engineering and applied mathematics, and I am very happy to see that here at Imperial, we have an interdisciplinary programme."
What are your future career goals and how have they been realised since being at Imperial?
Before Imperial, I was not planning to stay in academia after my PhD. The main reason was that I thought being in academia would make you a bit disconnected from the business world. But Imperial has changed my mind. It showed me that I can stay in academia and it can even boost my success in the business world. There are many professors who do research at the College and do a lot of consulting jobs for very well-known technology companies. This is optimal for me!
What advice would you give to someone considering applying for the Doctoral programme?
Find your ‘city’. A PhD is not something that you will work on 24 hours a day on and not go out. A PhD is something that makes you engaged with the city you live in. I was very afraid of the possibility that I would end up at a city which is not of my type. After coming to London, I noticed that this is where I should be. Every morning when I come to my office, I walk from Gloucester Road station and on the way, I grab a takeaway coffee, sip it and look around to see beautiful streets and buildings. This makes me feel alive. I wouldn’t do a PhD in a city which I don’t like. This is a very critical point, because I am aware of many people who apply to universities in cities which they do not have any idea about. So, my advice is to know the city where you will be living in. Ask students about their social lives rather than academic lives, because if you have a good work-life balance, you will be successful in your work anyways!
Related Blogs

Celebrating women at Imperial College Business School #BalanceforBetter
As champions of diversity and gender parity, we’re lucky to have so many incredible women at Imperial College Business School.

Powering the future generation of AI-led mutual funds
Recent alum Dr Yu Zheng has already put his Quantitative Finance PhD to good use by launching a FinTech start-up, Arraystream Technologies.

Alumni case study: Streamlining cross-border commerce
With PhD in Quantitative Finance and a background in international banking and trade, Dr Ramin Takin isn’t your ‘typical’ entrepreneur.