Accessability Links


Wednesday 28th February 2018
Tags: guest blog
Tell us about yourself...
I am a Senior Lecturer in Data Science in the Department of Applied Mathematics, Liverpool John Moores University. My expertise is in machine learning and statistical methods, and knowledge extraction from non-structured and extensive datasets. I have published 55 peer-reviewed manuscripts in areas such as machine learning, biostatistics, bioinformatics, biomedicine, and databases. I am originally from Colombia. I left my country around 14 years ago when moved to Spain to study a PhD in Artificial Intelligence.

What has your career path been?
I have worked in academia since I finished my undergrad studies. Soon after finalising my undergrad studies, I started as a lecturer in a Colombian University under a program for young academics. A few years later, I moved to Barcelona, Spain, to study my PhD and then came to the University of Manchester, after being awarded an EU fellowship to investigate cognitive models using EEG data. After having worked at Manchester in several research projects in neuroimaging, public health, and drug discovery, I accepted a lectureship position in health informatics (Keele University). After a year and a half, came back to Manchester, this time to MMU, as a Senior Lecturer in Data Science. Recently, I moved to Liverpool John Moores University to continue my academic career as Senior Lecturer in Data Science.

Harvard has described Data Science as 'the sexiest profession of the 21st Century', can you pin point the time when it became such an attractive industry to work in?
In the 90s and before, data mining, machine learning, and pattern recognition were topics mainly discussed in University’s classrooms. It was not until the mid 2000s when big companies such as Google and Amazon developed the infrastructure, that aforementioned disciplines became attractive and put them together as what we know today as Data Science. Since then, the number of companies doing data analysis or relying on data science has increased exponentially. Several factors have helped to this, but mainly because we have more powerful and cheaper computing systems now.

What sparked your interest in pursuing it further as a career?
When I started my PhD, I wanted to apply neural networks, which now happens to be one of the core methodologies in Data Science, for image processing and computer vision. Then I realised very quickly, that what I actually wanted to do was to develop data science methods rather than applying them. Since then, and even that I realise the importance of the application domain, always care about the methodological aspects of data science.

Your current role sees you as a University lecturer. What do you like about your role, what challenges do you face?
One of the aspects I like the most about my role is the possibility of being at the forefront of data science research. I also like teaching data science. That gives me the possibility of bringing real-world applications into the classroom and discussing them with the students.

In what way do you think the industry could support universities in creating the best Data Science graduates?
Currently, students enrolled in sandwich-format courses have the possibility of gaining experience through placements. We are convinced of the major positive impact on our students. However, we find that the number of places is relatively limited if we consider the size of the industry in the North-West. More companies need to realise the great mutual benefit of opening the doors to students.

What tools and technologies are the universities providing in order to equip students for a career in data science?
Undergraduate students within Computer Science and/or Mathematics departments are usually exposed to some general aspects of data science such as learning the data science cycle and associated programming languages (Python, R, etc). However, Universities are increasingly offering MScs, in which data science is studied in depth. Also, they are investing in high-performance computing infrastructure and platforms for big data analysis.

What are the key areas of data science you think is most important to teach data science students of 2018?
Rather than focusing on a specific aspect of data science, students should be able to understand the complete data analysis cycle, from exploring the data to presenting the outcomes. Also, I consider relevant to teach programming languages paradigms for data science, especially functional programming. I believe this experience gives the students the needed background to easily adapt and/or learn the specific technologies depending on the sector after the University.

Gender equality is something that is present in the tech and digital sectors. Do you see that there is a significant difference in numbers of males and females studying Data Science? If so what do you think could be done to bridge the gap?

Although I do not have precise numbers, I can see the difference in numbers is still significant. There are several reasons that can explain the current gender imbalance in data science jobs. The lack of STEM education is still more prevalent in women, for instance, but also, many companies tend to ignore government gender balance policies. Companies have to do more for bridging the gap. It is important for them to realise that numerous studies have shown that companies perform much better when their workplace is diverse. 

How do you keep up with trends in the industry? What do you think is next for data science in 2018?
My main source of information is scientific journals. I read lots of data science articles almost every day. Also, data science magazines are a very good source to keep me up with the trends.  I am also in permanent touch with companies and always have to keep an eye on job ads so I can pass the information to my students. Because of that, I am aware of trends in the industry, so I attempt to adapt my lectures accordingly.

What is next for data science? I think the next step is about interpretability. There is a general issue shared across many data models which is that they tend to provide answers, make predictions, but they offer little details about why they came up with those results. They are typically black boxes. I believe there is enough momentum now for moving towards more interpretable models, a sort of grey (hopefully white) boxes.

Add new comment
Back to Top