In the Spotlight: Héloïse Stevance

Dr Héloïse Stevance is a Schmidt AI in Science Fellow at the University of Oxford. In 2024 she was awarded the Caroline Herschel Prize Lectureship. Here, Héloïse talks about speed viewing Rubin images and shares the secret to doing robust science.

Can you sum up your work?

Find supernovae before they explode, make automated tools that help me along the way, and share my tools with the wider community so they find the best transients to answer their science questions. It's a mix of astronomy, computer science and machine learning.                 

What excites you most about the Rubin LSST?

The depth of field and the breadth of its community.

What inspires you to do the work you do?

I think the UK contribution to Rubin is well-regarded and significant. I think my work can go quite some way to maintaining and growing that reputation. I also look forward to seeing astronomers begin to work on the survey products, which are now just a few months away.        

Give us a surprising work-related fact.

As part of a Zooniverse project, I eyeballed 40k early Rubin triplets (reference, target, difference) to classify them as real or bogus. At my fastest I can do 50 per minute, that's 0.8 per second. If I had to do this for the estimated 10 million alerts per night it would take me 100 days (no sleep no breaks).

Now that's not even realistic since the 10 million are real, but if there are two bogus alerts for every real one (generous ratio), it would take me about a year! Thank God for computers.

What was your pathway to your current work?

I owe everything to the MPhys in Astronomy I did at the University of Sheffield (and the year I spent in La Palma during my Masters year).        

What's your favourite thing about being a scientist?

STARS GO BOOM! And then we can see them from so far away that the dinosaurs had not yet walked the Earth when we see them in LSST most of the time!     

What do you enjoy doing when you’re not working on LSST:UK

Cooking beans (I recommend buying the Bold Beans recipe book) and playing video games with my husband – I’m currently playing A Ship of Fools.

If you weren't a scientist, what would you be?

A volcanologist, so I can play with rocks and lava.

 Is there a book you’d recommend to anyone interested in astronomy?

The Accidental Universe by Chris Lintott is really good.

What's your advice for aspiring astronomers or scientists who wish to do the kind of work you do?

You can't out-compute bad/incomplete data. Just because you slap a fancy stats method or ML algorithm on something does not make it more ‘science-y’. Use the simplest method that will get you to a robust science answer.