EPCC is hiring data scientists

Our situation in the University makes us fairly unique. We have a diverse mix of both research and commercial projects so it is possible to work on developing the state of the art and also with some big names on work that has a real-world impact. Unlike many technical/research positions at the University, which are short-term Post-Docs, at EPCC our positions are longer term (initially a 2-year fixed-term contract) and many staff have been working at EPCC for several years.

One of our major strengths is our heritage in high performance computing (HPC), not least hosting the current UK national supercomputer ARCHER and many of the previous generation machines too.

HPC meets data analytics

I recently worked on a project with the UK Met Office where we combined HPC with data analytics. As some background to this we developed a new atmospheric model, called MONC, which is used by the UK weather & climate communities for modelling clouds. This became very interesting because scientists themselves are not particularly interested in the raw data (not least because it is generally huge) but instead in higher level information produced from it.

The data analytics techniques adopted by previous generations of atmospheric models, which themselves dealt with a few million grid points over maybe a thousand or so cores, were no longer applicable to the billions of grid points over tens (if not hundreds) or thousands of cores that our model was capable of handling. Therefore, to support the scientists, we had to develop a novel in-situ approach to data analytics. This is where cores of a processor are shared between the computational model and data analysis. Most often one core in a processor will be dedicated to data analytics, servicing the remaining cores which perform computation.

High performance data analytics (HPDA) is really interesting, and this combination of traditional high performance computing with data analytics will become far more common in the next few years and drive significant new advances. Existing HPC and data analytics techniques had to be specialised to support HPDA and with our approach we were able to show scaling to over 36,000 cores on ARCHER. For me this project was a nice mix of research, progressing the state of the art in HPDA, and impact, where scientists actually rely on our work and use it to further their understanding of the atmosphere and how clouds form.

Join EPCC, see the world!

Travelling around the world and meeting lots of different people is another aspect of working at EPCC. The project I mention here involved two trips to the US – the first to Salt Lake City in 2016 for SuperComputing (our largest conference) where I presented an early version of this work and then to Seattle in 2017 where I presented a more mature paper on HPDA. I find being able to see lots of new places and meeting a real diverse set of people really useful. It also gives a different dimension to what I am working on, as well as generating lots of new ideas and forming some really useful collaborations too.

The variety of work is something I think often attracts people to working here. I don’t just mean the variety of different projects (although we do have lots of interesting things to work on!) but also the other activities you can get involved in. From teaching on our MSc programmes to supervising students, from getting involved in one or more of the numerous outreach events we attend each year to helping write project proposals in areas that you are interested in – there is plenty to do in addition to core technical work.

Creating the data capital of Europe

There are also lots of new projects on the horizon. One of the drivers behind this is the Edinburgh City Deal, which aims to make Edinburgh the data science capital of Europe. As you can imagine EPCC is a key part of this and whilst the City Deal is just starting out, the next few years will see the design and delivery of the world-class data infrastructure. This will be hosted by EPCC and used to solve the next generation of data science problems. Driven by this we will also have many associated projects engaging with local, national and international organisations focussed around data science and HPC.

 So if you are interested in a new and exciting career, solving state-of-the-art challenges and being at the forefront of data science, then consider our vacancies. As I say the closing date for applications is the 22nd of February.