Centre stage with Tom Liptrot, Freelance Data Science Consultant at Ortom
Born – I was born in Orkney, an island in the far north of Scotland. My parents are sheep farmers there.
Studied / Education background – For my undergrad I studied Maths and Philosophy at Edinburgh University. I then did a graduate diploma in Economics at Birkbeck College London and an MSc in Applied Statistics and Operations Research at the University of Salford.
Current role / bio – I'm a freelance data science consultant. I help companies to solve problems and create opportunities using data and machine learning. I've been a data scientist since the term was first invented about 10 years ago and since then I've worked in a range of industries including healthcare, tech start-ups, pharmaceuticals, retail, and industrial. Before setting up my own consultancy I was head of data science at Peak.ai
What does your day to day role entail?
Most days I am working with clients on data science projects. I spend a lot of my time building machine learning models using R and Python. Usually this means trying to predict or classify something that will be useful to a business. It can be quite varied; examples include:
- classifying text messages as threatening or not,
- predicting which patients will respond to a drug treatment,
- predicting which job candidate will be most successful in a given role,
- predicting which sales leads are most likely to convert,
- forecasting daily demand for a product.
I also mentor data scientists and work with managers to develop and execute AI strategies.
What’s been your biggest work achievement of the last 12 months?
Setting up my own consultancy Ortom. It’s been hard work, but so far it’s gone pretty well! I’ve worked on some really interesting projects so far. I’m currently working with a start-up called Cogenis, where we have built an AI system that uses natural language processing on social media data to help adults keep their children safe online. I have also been working with a mobile phone repair company where I built a route optimisation solution that will reduce their weekly mileage by 10%.
Slightly differently, I also did a stand-up comedy show about data science last year in Edinburgh, which was a first (for me, not Edinburgh, although I don’t think there has been much data science comedy yet).
What is the biggest challenge facing the industry?
Avoiding another tech crash. Huge amounts of money are pouring into start-ups in the hope of finding the next Google or Facebook, but the failure of recent IPOs from Uber and Wework show that even the most successful tech companies are struggling to become profitable. A slight change in expectation of return could see money drying up. How that would play out given a lot of the exposure is private, remains to be seen. Also, the domination of the big tech firms has the potential to be unhealthy.
What are the top 3 challenges facing your organisation personally?
Combining all the multiple roles needed on running my own business - I’m now the finance director, head of sales and marketing as well as doing all the data science stuff! Keeping up to date with all the latest developments in AI. There are lots, choosing which ones to learn about is hard, as I would love to learn them all but don’t have the time. Keeping focus, when there are so many options and opportunities.
What’s the best piece of advice you have ever been given?
Probably my statistics lecturer (Prof Ian McHale) advising me to learn R in 2006.
What are your predictions for the IT industry for 2020/21 or beyond?
As NIels Bohr may have said, “Prediction is very difficult, especially if it's about the future”. I’d guess, more data, more concerns about privacy, more developments in AI.
Why do you think everybody is talking about AI being important for digital transformation but companies are still reluctant to invest?
The impact has not been shown widely. The technology is young. Companies don’t have expertise in using these technologies, or in hiring and managing people who can. AI is a new type of disruptive innovation and companies are taking time to adjust.
The idea of the productivity paradox: the time lag between a technology being introduced and seeing productivity gains – is a well-known phenomenon. Technology enters industries and often those industries don’t quite know what to do with it. Often that is where I can be helpful - helping people see what is (and is not) possible. If you try to solve the wrong problems with AI it won't work. This is what my talk is about at the Digital Transformation EXPO.
What do you think is going to be the next big technology development?
In AI, In the last couple of years there have been some really big developments in language based AI due to the large transformer based neural networks. These are models that basically try to predict the next word in a sentence (like predictive text). They are trained on very large amounts of text (Wikipedia and thousands of books). The new techniques that have been developed have pushed forward the state of the art performance on a range of tasks including, translation, summarisation, text generation, question answering, chat, and text classification. I would expect these developments to start coming through in commercial products and applications.
There's also been a lot of developments in model-based reinforcement learning. This is an approach for creating agents that can learn from experience by creating models of the world. Using a model rather than a model-free approach can reduce the amount of data needed to train an effective system.
It is thought this type approach is more similar to the way humans think. This is still fairly theoretical but could have a huge impact on the applications of AI as it can improve the ability of a model to learn. In general, the ability of AI to work with less data seems to be coming.
Do you think GDPR has impacted your role in a big way since its introduction?
For me personally GDPR hasn't had a massive impact and I think some of the companies that I work with have been impacted, mostly around the management of data and the right to deletion. Most of the rules GDPR brought in, should have been followed already, the main significant change was the size of the punishment if you break them.
What does digital transformation mean to you – what in your opinion is most important to a successful implementation?
Digital transformation to me means using technology to improve a business. To do that well you have to understand the business and what needs and requirements are. In data science and AI, that means understanding how the data relates to the business, where the data comes from, who creates it and who uses it.
I think the biggest problems come when people try to impose solutions without there being a desire or even a need for them. Getting the people who will use a solution on board is key.