- By Analytics Engines
With this year’s Big Data Belfast conference just around the corner, the team at Analytics Engines is profiling some of the event's most notable speakers.
It might seem an unusual jump to go from nuclear engineering to data science, but Dr Clair Sullivan says the two disciplines have more in common than you might realise. In a fascinating podcast interview ahead of her keynote talk at Big Data Belfast, the pioneering machine learning engineer reveals how she made the leap from one discipline to the other and how her diverse team at software development platform GitHub includes astrophysicists, educational experts and even a lawyer.
“I actually started my career at Los Alamos National Laboratory in New Mexico and that was as a nuclear scientist. That was 2002, so you're talking a year after 9/11 - and 9/11 had a profound impact on me,” she says. “I wanted to do what I could for the world, to make the world a safer place, working on things like nuclear emergency response. But the interesting thing is that I was working on maths problems using techniques that are actually the basis of some deep learning today.”
She returned to academic research in 2012 as an assistant professor in the Department of Nuclear, Plasma, and Radiological Engineering at the University of Illinois at Urbana-Champaign, and that was when the data science bug bit. “I’d been working in basically the combination of the Internet of Things with nuclear engineering, trying to make sense of huge networks of sensors to find the illicit movement of nuclear material. And it occurred to me well there’s a lot of techniques here that I don't know, I was never trained in,” she says.
“I got into this boot camp called the Data Incubator, a seven-week programme to take people with degrees in things like physics and maths and turn them into data scientists. And I loved it, I absolutely loved it. “It really inspired me, looking at all these people coming from different backgrounds and applying the same skills. The math skills that I learned as a physicist could also apply in this really exciting and evolving world.”
Dr Sullivan describes a student who applied their work with bubble detectors to convolutional neural networks and ended up working at Facebook. Meanwhile, she cites time series analysis - widely used across the fintech sector – as an important tool in nuclear engineering.
She explains: “We were doing time series analysis on some of our data on our sensor networks… One access that we looked at was the energy of the particles, but the data is identical to time series data. So the same maths that applied to what we were doing in nuclear also can apply to fintech or business analytics - there’s a lot of parallels there!”
Part of the work that GitHub is doing lends itself to tracking programming languages over time, watching languages like Python grow and evolve, while others like Fortran 77 decline. “The growth of Typescript is probably one of the bigger growths going on right now… and certainly you see Javascript, Python,” Dr Sullivan says. “Obviously Ruby on Rails will always be a language that is very near and dear to us, because that’s what GitHub is based on, so Rails is important to us.”
Clair Sullivan is speaking at Big Data Belfast on Thursday October 24 at the ICC Waterfront. For more information and to book tickets visit www.bigdatabelfast.com. Listen to Clair's full interview on the latest Analytics Engines Insights podcast.
Follow @BigDataBelfast and @AnalyticsEng on Twitter for more information.
Presented by Analytics Engines and supported by headline sponsor Ernst & Young (EY), supporting sponsors Dell, CME Group, Allstate Northern Ireland, SmashFly, SpotX and Altra Executives and industry sponsors Invest Northern Ireland and Signifyd, the event promises to provide insight into emerging trends and issues in data and business analytics.
