From Applied Machine Learning to PROWLER.io — Simon Fothergill on his Data Science career
“STEM is about problem solving: If you don’t know what that is, find out. If you don’t know how to do that, find out.” — Simon Fothergill
What does it mean to be a data scientist?
It is an applied research position. Data scientists find out the questions customers have about their areas of interest (e.g. smart cities or financial trading) by talking to them. We then look for general ways of applying tools, that other “theoretical” researchers have discovered, to the clients’ data in order to answer these specific questions. Different data scientists might have different areas of focus, but there may be some commonality to the questions they focus on, which could then be distilled into general concepts about life: discovering ways of using algorithms that take advantage of these general concepts is the Holy Grail of AI. There are also theoretical researchers that come up with new algorithms (tools), as well as software engineers that build different tools to support the researchers’ experiments.
Where do you currently work?
I am working at PROWLER.io in Cambridge. It is an AI company, a two-year old start-up that has grown really rapidly: there are already 80–90 people working there. PROWLER.io is building the latest AI platform that automatically makes decisions. The thing that makes it special is that it is a nexus of different research communities that haven’t worked together before. Probabilistic Modelling, Reinforcement Learning and Multi-agent Systems are the three pillars of the business.
Can you tell us more about the three pillars of PROWLER.io?
Probabilistic Modelling is about machines learning from examples given by humans. Such as, telling the machine that this photo is probably of a dog, there is probably a person in front of the car now, something will probably happen next as it did last time we experienced this situation. Probability is a good way to deal with the uncertainty in the world, when making decisions.
Reinforcement Learning is akin to psychological conditioning (but of machines!) through positive or negative reinforcement of behaviours. It is about the machine learning the best plans of action.
Multi-agent Systems considers what actions will be carried out by a group of machines and what individual machines should do in certain situations, where they are working together.
How did you get your current job? What was the process?
I saw PROWLER.io on the internet advertising positions and I was talking to recruiters. As it happens, the CEO of my previous employer helped us to secure new positions through her links to PROWLER.io and arranged for a meeting with a recruitment contact at the company. I sent my CV, attended an interview and then they gave me a coding problem to solve. That is a pretty standard process in recruitment for software engineering roles and data science positions. I had to demonstrate mathematical knowledge, coding skills, architecture skills, algorithm understanding and whether my personality fit the team.
What first attracted you to a career in computing?
I guess it all started because my dad was head of computing at a school and in the holidays he had to go into school to work on the systems; my brother and I had to go with him and to keep us occupied we played with computers. I was about 9 or 10 years-old, when I got into programming. My first language was RM-Logo and I started making funny little games. I am very analytical and I like building things but I also like getting “inert” machines to do something by themselves — that is quite cool.
What qualifications do you need to do your job?
I studied mathematics, physics, chemistry and computing at A-level and then decided to study a Computing degree run by the Faculty of Engineering at Imperial: it was a four-year course, which included industry experience. I wanted to make machines do things that humans can do, but better. I was keen to keep learning so I went onto a Ph. D. in Computer Science at Cambridge, which was more of an education into science that became an applied machine learning Ph.D. It was natural to move into start-ups that enabled me to continue to pursue AI computing. Data science is applied AI research in data analytics. People have always needed data analysis and I thought they always will; AI research contains some of the most advanced analytical approaches that exist so it is a good thing to study. At the moment, computing power and availability of data are fueling AI research.
What does your job entail day-to-day?
I sit at a desk but there is a lot of walking around the office and nearby gardens! There is good opportunity to network and explore other parts of the business and it is a company built on research. There are no real ‘office hours’ but we are contracted to a standard 37 ½ hour week. There are lots of talks and we are encouraged to connect with people all over the company. At the moment, I am learning the ropes through completing ‘tickets’ at a software engineering level, which supports other teams that conduct research and build products.
Can you describe any challenges or triumphs in your career, so far?
Personally, as a curious and analytical person, I have had to learn to focus on what is said, what I say and what I do. The biggest triumph’s so far are having my Ph.D. approved and getting my latest job offer.
What advice do you have for someone starting out in STEM?
Make a contribution. Be original. Do a Ph.D. because it develops your confidence as an individual thinker, your professional stamina, your imagination and your problem solving skills. STEM is about problem solving: If you don’t know what that is, find out. If you don’t know how to do that, find out. Always be aware of what is coming next.
If you could meet anyone STEM-related, who would it be and why?
Albert Einstein, as he would have had to have come forward in time and he would have liked to have done that!
Where do you see yourself in five years’ time? What are you planning to do next?
I see myself directing contributions to knowledge, as a more responsible and experienced, leading data scientist.
“I wanted to make machines do things that humans can do, but better.”
— Simon Fothergill
Oxbridge Inspire delivers innovative STEM education and provides guidance and inspiration to young people wishing to pursue STEM subjects at University and beyond. To find out more about Oxbridge Inspire and the courses and activities we offer, visit our website.
