Computational scientist Elise Jennings talks about her career journey from astrophysics to supercomputing and how the STEM industry has changed in recent years.

The Irish Centre for High-End Computing (ICHEC) recently appointed Elise Jennings as senior computational scientist. Jennings previously worked at the leadership computing facility at Argonne National Laboratory and as an associate fellow at the Kavli Institute for Cosmological Physics at the University of Chicago. She has a research masters degree in theoretical physics and completed a PhD at the Institute for Computational Cosmology at Durham University.

Unsurprisingly, her passion for STEM subjects started young. When I was studying for the Leaving Cert, I loved biology, mathematics and applied maths. Solving science problems and finding out how things worked was a lot of fun for me, she said.

I especially loved those rare moments where I felt I had understood something at a deep level or when a simple event in everyday life can lead to some interesting mathematics. For me, the fact that we could describe these events precisely and make predictions was amazing and certainly got me hooked on science.

The adoption of machine learning and deep learning methods for science has been phenomenal and has enormous potential to accelerate scientific discovery ELISE JENNINGS

Although she started her research work with a focus on astrophysics, Jennings said moving to supercomputing was a very natural transition because her PhD was focused on modified gravity and testing those theories using large N-body simulations of structure formation in the universe.

These simulations ran on a dedicated cluster at the Institute for Computational Cosmology at Durham University and this gave me a great opportunity to understand how these codes run at scale and what it means to profile code for performance, she said.

Overall, many scientific applications from different domains require HPC [high-performance computing] systems to handle the data velocity or size, or to solve the complex scientific questions using advanced methods and simulations. My work has now expanded from astrophysics to include many of these areas and work with researchers looking to scale up and optimise their codes.

In her new role with the ICHEC, Jennings will be responsible for the creation of an exascaling team as part of the European High-Performance Computing (EuroHPC) Competence Centre for Ireland. This centre will operate a programme of mentoring and upskilling the most ambitious and scientifically accomplished academic groups in Ireland.

Jennings said this will enable Irish researchers to migrate from the national Tier-1 system to EuroHPC Tier-0 supercomputers, in preparation for European exascale systems and beyond.

It is exciting work engaging Irish research groups and industries as they develop competitive proposals for EuroHPC resources. I will also be developing HPC training programmes for simulations and emerging scientific machine learning and deep learning methods.

Jennings said one of the challenges she has encountered in her career is academic job uncertainty, noting that the first hurdle after a PhD is securing a postdoc, which can be very competitive.

After a couple of postdocs, the pressure is on to find a permanent research position or lectureship if you want to stay in academia, she said. After dedicating so many years to studying astrophysics, it was difficult to face the prospect of leaving and this fear would arise every couple of years when I had to reapply again. Tied in with this was the pressure to publish, which can drive research on but also adds stress if a line of research is not productive.

However, she also spoke about how incredibly lucky she feels to have worked in some of the top institutions in the world, witnessing cutting-edge research in real time.

From taking part in large astrophysics collaborations at Durham, [University of] Chicago and Fermilab, to working towards the first exascale machine in the world, the Aurora A21 machine at Argonne, she said.

A recent highlight that stands out for me was taking part in a small research team running benchmarks on a dedicated AI testbed at Argonne. I was one of the first people to run a deep learning benchmark on the Cerebras wafer-scale AI chip. It was very exciting to take part in cutting-edge development and innovation like that.

Jennings added that she has noticed several changes in the STEM industry since she began her PhD, including the increased need for HPC systems to process data or run simulations within scientific applications.

The adoption of machine learning and deep learning methods for science has been phenomenal and has enormous potential to accelerate scientific discovery. Many of these methods are built on advanced statistical techniques such as regression, which have a long history, she said.

Deep learning is a new methodology that allows us to process and understand very diverse datasets. Deep learning also poses some new challenges for HPC, particularly in terms of creating AI-driven infrastructure and code, handling massive datasets and coupling with traditional simulation and modelling techniques.

From a broader angle, Jennings also said the increase in diversity at all levels of STEM is another welcome change, which is slowly changing the perception of what it means to be a scientist.

At every stage of my career, I was very aware of diversity in my research groups and how it impacted the culture and productivity. It is encouraging to see more and more conversations about this and hopefully it will continue to bring about positive changes at all levels.

Jennings added that she would advise women who are pursuing a career in computer science to seek out companies and institutions that have a good reputation for a diverse workforce and a healthy work culture.

Most companies now recognise how crucial work-life balance is for the happiness and success of its employees but some academic institutions have been slow to learn these lessons, she said.

I would also recommend to any woman interested in pursuing a career in computer science to connect with the Women in STEM groups or outreach events which take place at their institution. It is invaluable to make contacts with leaders in the field at these events and hear their perspectives.

Want stories like this and more direct to your inbox? Sign up forTech Trends, Silicon Republics weekly digest of need-to-know tech news.

Excerpt from:

'Moving from astrophysics to supercomputing was a natural transition for me' - Siliconrepublic.com