I'm an Australian Deep Learning Postdoctoral Research Fellow working from the University of Melbourne. My research interests span from adversarial risks against Artificial Intelligence to Reinforcement Learning, Graph Neural Networks, Game Theory, Nonlinear Differential Equations, and Fluid Dynamics. This website was coded entirely using ChatGPT, without writing a single line of code (for more details click here).

Outside of research, I value mathematical creativity and kindness, and enjoy hiking, cycling, a wide range of music and weightlifting.

Contact me here to discuss Deep Learning or AI security, or topics in Machine Learning more broadly.

About me

A research scientist sitting at the intersection between deep learning and mathematics, I attempt to develop insights into systems using a mixture of analytical analysis and numerical computing.

  • Location:
    Melbourne, Australia
  • Age:
  • Nationality:
  • Interests:
    Cycling, Running, Weightlifting, Formula 1, Weird Fiction
  • Study:
    Monash University
  • Employment:
    University of Melbourne


Doctor of Philosophy (Science)

Monash University2012 -- 2018

Developed a fast, scalable and spectrally accurate numerical technique for solving nonlinear variable coefficients boundary value problems, and applying them to atmospheric wave dynamics. Beyond this I also worked with applying gradient-free optimisation techniques to solve applied mathematics problems traditionally solved using brute force techniques. Over the course of my PhD I also consulted on the application of mathematical techniques to Monash Laboratory for Optics & Applied Mathematics and the Laboratory for Turbulence Research and Combustion.

Bachelor of Aerospace Engineering & Bachelor of Science

Monash University2007 - 2015

Developing a unique compressed sensing based closure for the simulation of turbulence in Large Eddy Simulations and modelled structural and flight characteristics of a waterbombing aicraft. Using Geometric Programming and a Genetic Algorithm to optimise the equations led to a 50% reduced mass compared to conventional techniques, with a calculation time of 5 seconds.

Bachelor of Science (Honours, First Class)

Monash University2011

Thesis: Contour Dynamics within an Annular domain. Graduated at top of class with H1 (First Class) honours. Also attended the Australian Mathematical Sciences Institute Honours school through The University of Adelaide, receiving a HD average.


DAAD-Ai Fellow


Selected as a 2023 DAAD Ai-Net Fellow for 2023.

Postdoctoral Research Scientist

University of MelbourneJune 2019 - Present

My work builds upon a core idea: how mathematical tools can be used to glean better insights into complex systems. This includes coming up with novel approaches for considering both adversarial risk and defences in deployed machine learning systems, and work that sits at the intersection between game theory and complex nonlinear differential equations.

As part of this I am currently the project lead on a $450,000+ grant covering Certified Defences against models other than classifiers; and the research lead on two additional projects, one covering adversarial behaviours against classifiers, and the other on game-theoretic solutions to complex organisational problems.

I also spent 2 years as the research lead on a large, inter-disciplinary applied project covering the modeling, simulation, and prediction of Wireless spectrum. Over the course of this project I regularly worked with C-suite executives to provide technical expertise (including dozens of technical reports) on both modeling and managing very large data sets.

Outside of my research work, I regularly provide technical consultations to other researchers on both deep learning and how to leverage large-scale computing resources to improve research outcomes.

Teaching: Regular guest lecturing roles on graph neural networks, security, and distributed systems, and part of the team designing and lecturing a new Security Analytics Masters subject at the University of Melbourne.

Supervised PhD students: Marc Katzef (Graduated, 2023), Ifrah Saeed, Shijie Liu, Takuma Adams, Tiancheng Jiang.

Collaborators: Benjamin Rubinstein, Tansu Alpcan, Sarah Erfani, Christopher Leckie, Ella Alipourchavary (UoM), Paul Montague, Alex Kalloniatis (DST) and Justin Kopacz (Northrop Grumman).

Other activities: Member of the Security Group within the University of Melbourne School of Computing & Information Systems (CIS); TRAMx Research Entrepreneurship Participant (Best Pitch, 2023); TRAM Tracks: Project leader on developing Secure Database Services; Regular consultant on GPU-Enabled HPC within CIS; external expert on hiring AI/ML talent for DST Group Australia.


Monash UniversityMarch 2009 - October 2018

Tutoring a range of Engineering and Mathematics subjects, as well as serving as a live, text-based assistant lecturer for online lecturing, and specialist one-on-one tutoring to disadvantaged students.

As head tutor, I was also responsible for preparing materials to support a large team of other tutors.

Regularly received student appraisals above 90% positive.

Consultant Mathematician and Data Scientist

Self EmployedMarch 2007 - June 2019

Providing insights and tooling for a range of clients, including large ASX listed companies and sporting leagues. Projects included tracing millions of lost chargeback revenue; designing a dynamic bid-ranking algorithm for a tech company with hundreds of thousands of active users; a court allocation and scheduling program for a netball league; and providing advice for the Victorian Emergency Services games regarding algorithmic fairness.

Mathematical Modeller (Sports)

Self EmployedMarch 2012--October 2017

Developing models that captured significant value for betting on the Australian Football League, while also pursuing arbitrarge opportunities across a range of sports. Project ended after being banned by all major Australian bookmakers at the time.


Technical Skills
Deep Learning
Numerical Analysis
Mathematical Modelling
Data Analysis
Algorithm Design
Personal Skills
Technical Writing
Project Management


LightEM (Light Experiment Management) [Coming Soon]

A lightweight experiment management system that requires no databases, no installs beyond a single Python library. Designed for experiment management from small scale to huge, tracking, and analysis on systems where loading a webserver or establishing a permanently running database is infeasible.
On launch LightEM will also provide an optional website for graphical analysis, although all data will be retained locally in an easy to parse fashion.

Coding with chatGPT

A website where the (majority of the) code was written exclusively using chatGPT.

Abridged Bibliography

    Recent Posts

    Code free Coding with chatGPT
    © Copyright 2023 Andrew Cullen