Skip to content
STARS / SPECTRA / MODELSSHEET 01 — ABOUT

Computational
AstrophysicistBuilding AI-Powered Toolkits

A rotating star and an orbiting exoplanet. Explore my workPublications 6
Star + exoplanet01 / 06

Transit spectroscopy / secondary eclipse / thermal phase

Illustrative models.

SYSTEM / IN MOTION
University of Cambridge Research Assistant Institute of Astronomy Explore my workPublications 6
  • Data Scientist
  • Machine Learning Engineer
  • AI Researcher

My research turns astronomical spectra into constraints on the physics of complex astrophysical systems. At Cambridge, I work with Nikku Madhusudhan’s Hycean group to disentangle planetary signatures from the changing spectra of their host stars, bringing together stellar physics, machine learning and Bayesian inference for JWST observations.

Previously, I developed SPECULATE during my doctoral research at Southampton: an open-source framework for emulating accretion-disc spectra and probing the physics of their winds. My work has also taken me beyond astronomy, into numerical weather prediction at IBM Research and fusion-plasma diagnostics at UKAEA.

Publications 6

  1. Optical outburst evolution of the transient black hole X-ray binary Swift J1727.8−1613: disc response to jet ejections and late-outburst emergence of powerful disc windsMNRAS
  2. Radiation-ionization hydrodynamic simulations of AGN line-driven winds lead to transient shielding and BAL/UFO signaturesMNRAS
  3. The critical role of clumping in line-driven disc windsMNRAS
  4. A diagnostic kit for optical emission lines shaped by accretion disc windsMNRAS
  5. Monte Carlo radiation hydrodynamic simulations of line-driven disc winds: relaxing the isothermal approximationMNRAS
  6. SIROCCO: a publicly available Monte Carlo ionization and radiative transfer code for astrophysical outflowsMNRAS
7 ORCID works, including software

Places

From observatory nights to conference stages.

Group photo in Southampton
Doctoral research

Southampton, UK

Southampton colleagues during my doctoral research.

Austen Wallis in an observatory in Tenerife
Observatory work

Tenerife, Spain

Observatory work with the IAC and University of La Laguna.

Austen Wallis at Culham
UKAEA placement

Culham, UK

Plasma-pressure surrogate modelling at UKAEA.

Austen Wallis at the Hartree Centre in Daresbury
IBM Research placement

Daresbury, UK

Machine learning and optimisation at IBM Research’s Hartree Centre.

Austen Wallis speaking at Trinity College Dublin
Talk

Trinity College Dublin

Presenting WRF Weather Forge and global forecasting work.

Architectural rendering of a modern house in Winchester
Local work

Winchester, UK

Planning and development work with Wallco, alongside my PhD.

Austen Wallis speaking at PyData Southampton
Community talk

PyData Southampton

Machine learning and scientific computing with the local PyData community.

Audience view during Austen Wallis's PyData London 2024 lightning talk
Lightning talk

PyData London 2024

A lightning talk at PyData London 2024.

Austen Wallis receiving an award at GRADnet 2024
Award

GRADnet Award

Best PhD Poster at the SEPnet Employers Exhibition, 2024.

Group photo at GRADnet 2024
Network event

GRADnet 2024

The graduate-research community at GRADnet 2024.

Austen Wallis at STEM for Britain 2026
Showcase

STEM for Britain 2026

Presenting research at the UK Parliament.

Austen Wallis receiving the STAG Research Prize
Award

STAG Research Prize

Best Publication in Astrophysics, 2025.

Austen Wallis receiving the FEPS 3 Minute Thesis runner-up prize
Award talk

FEPS 3MT Runner-Up

FEPS 3 Minute Thesis runner-up: “Accelerating Astronomical Discoveries with AI Emulators”.

Curriculum vitae

Download CV PDF

Research & industry

  1. Sep 2026 — present

    Research Assistant in Exoplanetary Remote Sensing and Data Science

    Institute of Astronomy, University of Cambridge

    Modelling M-dwarf spectra and stellar variability to support exoplanet-atmosphere inference with JWST observations.

  2. Sep 2022 — present

    Postgraduate Researcher

    University of Southampton

    Developing SPECULATE, simulation emulators and spectral diagnostics for accretion outflows as part of my doctoral research.

  3. Jun — Sep 2025

    AI Research Scientist

    IBM Research, Daresbury

    Non-gradient optimisation in PDE solution spaces and synthetic weather data for the IBM–NASA Prithvi effort.

  4. Jun — Sep 2024

    Machine Learning Research Scientist

    UK Atomic Energy Authority, Culham

    A deep-ensemble surrogate for plasma-pressure prediction, combining fast predictions with uncertainty estimates.

  5. Jan — Jun 2019

    Student Researcher

    Instituto de Astrofísica de Canarias & University of La Laguna

    Studying Jupiter's mass and rotational flattening, with observatory experience on the IAC-80, TCS and Mons telescopes.

Education

  1. Sep 2022 — present

    PhD candidate in Astrophysics

    University of Southampton

    Thesis: Emulator-based inference of imperfect and complex astrophysical models: Reducing Millions of Computational Hours to Minutes.

  2. Sep 2017 — Jul 2021

    MPhys in Astrophysics with a Year of Research

    University of Southampton · first class honours

    Final-year research on calibrating a cosmological standard candle using active galactic nuclei.

Teaching & talks

  • PHYS2022: scientific Python and data analysis.
  • PHYS2011: astronomy design and observation.
  • Southampton Astrodome outreach.
  • Talks at RAS ML/AI Astronomy, PyData Southampton, PyData London and Trinity College Dublin.

Selected awards

  • 2026 FEPS 3 Minute Thesis runner-up.
  • 2025 STAG Research Prize for Best Publication in Astrophysics.
  • 2024 Best PhD Poster, SEPnet Employers Exhibition.
  • 2024 Second place, DISCnet LIV.INNO ML Kaggle competition.

Technical toolkit

Modelling
Emulators, Bayesian inference, Gaussian processes, ensembles, MCMC and time series.
Code
Python, C and Julia; PyTorch, TensorFlow, JAX, NumPy and SciPy.
Data
Pandas, Polars, Dask, Xarray, Matplotlib and Vega-Altair.
Workflow
Linux, Git, SLURM, LSF, Jupyter, Marimo and Streamlit.

Browse publications