Gareth Kennedy

data scientist and mathematical modeller


About me

I'm a data scientist with over a decade’s experience in mathematical modelling (aka predictive analytics), computational physics, data mining, statistical analysis and high-performance computing. Since returning from China I've been running a data science consulting company based in Melbourne Australia.

Before returning to Melbourne, I was a data scientist at MCON group based in Shanghai where I was responsible for developing machine learning solutions and predictive analysis models primarily for our automotive clients in China. I typically write in R for statistical solutions and Python for image manipulation applications and deep neural network applications (with TensorFlow and Keras).

Before working as a data scientist I worked in astrophysics, electrochemistry and geophysics, building computational mathematical models of complex systems, often from scratch using the fundamental physical equations. This has been conducted in three countries, namely Australia, China and Spain.

My resume can be found here, LinkedIn profile here and I can be contacted via email at: gareth.f.kennedy -at-

Photo taken at a TEDx talk I gave in April 2015 in Beijing entitled "Chaos. Order. Beauty".

Summary of background

Consulting Data Scientist: 10/2017 – present
Established a consulting business to provide data science, statistical and predictive analytics services. Academic clients include the University of Oxford, Monash University and the University of York. Current services provided include:
• Building and deploying Deep Neural Networks for automatic classification of experimental results
• Communicating results via peer-review papers
• Applying Bayesian statistical techniques to data; e.g. for electrochemistry using a Dockerised suite
  of Jupyter notebooks available publicly via github
• Providing predictive analytics and statistical solutions to clients

Data Scientist: MCON Group
Shanghai, China (10/2016 – 10/2018)
Highlights included:
• developing purpose built Deep Neural Networks for object identification and damage classification from vehicle image data
• designed stratified control groups and provided automatic AB testing for a large German automotive company in the Chinese market
• deployed predictive BI products using Docker on either the clients’ own servers or AWS

Postdoctoral researcher, School of Earth Sciences, University of Melbourne
Melbourne, Australia (5/2015 - 6/2016)
Worked as an applied mathematician/software developer hybrid using python to further develop the ''underworld'' geodynamics modelling software.
The code for this project is open source and can be found on github and the webpage is here

CAS Postdoctoral Research Fellow (Astrophysics), National Astronomical Observatories of China, Chinese Academy of Sciences (NAOC/CAS)
Beijing, China (7/2012 - 1/2015)
During this time I was applying my skills in mathematical modelling and computational astrophysics to studying the galactic centre as part of the Silk Road project. More on the interaction between stellar dynamics and accretion disks in galactic centres can be found (here) and on applications of the stability of the three-body problem (here).
Some recent publications on these topics can be found here. This was an extremely interesting time, not just for the research, but also to be in Beijing during a period of rapid economic growth and social change. The everyday problem of severe air pollution in Beijing was a continuous source of conversation and inspiration for this project.

Postdoctoral Researcher (Mathematical Modelling), School of Chemistry, Monash University
Melbourne, Australia (1/2011 - 5/2012)
For a summary of my research during this time see this link. The most important achievement during this time was the development of the MECSim software program which models a wide range of electrochemistry problems and is freely downloadable here.
Update: the paper coming from the updated version of this code is now accepted and will be available in January 2017.

Postdoctoral Researcher (Astrophysics), Institute of Cosmos Sciences (ICC), University of Barcelona
Barcelona, Spain (12/2008 - 12/2010)
Worked on computational modelling of galactic centres using a purpose built parallel code that included the gas and stellar dynamics. As I arrived in Spain just in time for the global financial crisis, this was a very interesting time...

Melbourne, Australia (<2010)
• Graduated with a Ph.D. in 2010 (submitted late 2008, accepted early 2009) entitled
  "Problems in stellar and planetary dynamics", under the supervision of Dr. Rosemary Mardling at the
  Centre for Stellar and Planetary Astrophysics (now Monash Centre for Astrophysics) at the
  School of Mathematical Sciences, Monash University, Melbourne, Australia. My thesis is available as a
  summary and in its complete form here.
• Some previous examples of pursuits outside work places include: politics and paintings.

Back to top