Yoni Nazarathy
Mathematical Data Science Researcher, Practitioner, and Educator
y.nazarathy@uq.edu.au
CV (Academic) 
CV (Industry) 
Google Scholar 
ORCID 
@ynazarathy (Twitter) 
YouTube (One on Epsilon) 
GitHub 
LinkedIn
I am an Associate Professor at the School of Mathematics and Physics of The University of Queensland for part of my time and working in industry in the rest of the time as a consultant via my codirected company, Accumulation Point.
My academic expertise is in Machine Learning, Applied Probability, Statistics, Operations Research, Simulation, Scientific Computing, Control Theory, Queueing Theory, Scheduling, and Mathematical Education.
My industry consulting expertise aligns with some of my academic expertise and in addition includes areas such as software for large language models (LLM), general software development, and statistics of bioequivalence analysis.
Application areas of my work include epidemics, wireless communication networks, biostatistics, agriculture, power systems, software and hardware design, manufacturing logistics, healthcare logistics, engineering, fire detection, pharmacology, and road traffic networks.
See the full list of my
research publications. You may also find some of my
blog posts interesting, as well as
LLM related blog posts within Accumulation Point. Further lists that maybe of interest are
media coverage of my work,
software that I cocreated, and a collection
presentations that I presented over the years. Finally, here is a list of academic
funding sources for my work.
Particular recent highlights of are my coauthored books,
Statistics with Julia: Fundamentals for Data Science, Machine Learning and Artificial Intelligence, and
Mathematical Engineering of Deep Learning.
These are additional notable themes of my work:

Safe Blues  A method for control and estimation in epidemics  This is a COVID19 inspired project dealing with collection of additional information via virtual safe virus spread and using the information to make better predictions and control decisions. I lead this project.
[More...]

Reinforcement Learning of Restless Bandits  The problem of dynamically allocating resources out of a limited pool to an evolving project is a central problem in machine learning and operations research. This research theme focuses on one useful abstraction in this domain, called restless bandits.
[More...]

The Mathematical Engineering of Deep Learning  Deep learning is in many ways "mathematical engineering". It is a collection of methods that work well in practice to create an "Artificial Intelligence" illusion.
[More...]

The Julia Language for Statistics and Machine Learning  Julia is an exciting new language that is perhaps destined to be the data science language of the future.
[More...]

Structured Markov Chains and Linear Control Theory  The interface of Matrix Analytic Models (MAM) and linear control theory is interesting and nontrivial. The underlying mathematics for linear systems and phase type distributions is very similar and there is room to make connections between the two fields.
[More...]

The BRAVO Effect in queues  BRAVO stands for "Balancing Reduces Asymptotic Variance of Outputs". It is a queueing models phenomenon that I discovered. Understanding it requires understanding how queueing models operate at threshold regimes.
[More...]

Queueing Network scheduling and control  This theme deals with the analysis and control for scheduling decisions of complex queueing networks where items traverse different stations over time and resource allocation decisions are to be made online
[More...]

Parameter estimation in queues  This theme deals with parameter and state uncertainty in queues.
[More...]

Truncated distribution  This theme deals with univariate and multivariate distributions that are truncated to a particular region. Moment matching and truncation correction techniques are developed and implemented
[More...]

One on Epsilon and Mathematics Education  One on Epsilon is a mathematics education project that pulls resources from the web to aid school teachers, parents, and students better understand the "why" beyond just the "how" of mathematics. Additional mathematics education projects around this theme involve educating school teachers about the mathematics of "Artificial Intelligence".
[More...]
Courses and Workshops
I have been engaged with teaching courses and running workshops since 1999. See the
full course list.
My teaching is at The University of Queensland, in previous institutions, via the Australian Mathematical Sciences Institution, the Statistical Society of Australia, as well as other settings. Here are my main areas of teaching:

Mathematics for Data Science  This suite of courses deal with the mathematical background needed for understanding datascience, statistics, and machine learning algorithms. This includes, linear algebra, calculus, discrete mathematics, and probability.
[More...]

Data Science and Statistics Theory and Practice  This suite of courses includes core data science, machine learning including deep learning, and statistical inference courses.
[More...]

Probability and Stochastic Processes  This suite of courses include probability theory, the theory of stochastic processes, advanced courses on stochastic processes, and the applications of stochastic processes to operations research.
[More...]

Control Theory  This suite of courses deal with linear control theory, Markov decision processes, reinforcement learning, and other aspects of mathematical control and optimization.
[More...]

Introduction to Statistics  This suite of courses deal with introductory statistics for data scientists, engineers, and the general sciences.
[More...]

Software Development and Scientific Computing  This suite of courses deal with scientific computing, software development, and multiple aspects of programming.
[More...]

Industry Based Projects  This suite of courses deal with the supervision of students as they undertake industry projects in data science, statistics, applied mathematics, or engineering.
[More...]

Mathematics Education  This suite of courses deal with aspects of mathematics education, including professional development for school teachers.
[More...]
Research Students
Current Research Students: Clayton Forknall (PhD), Zhihao Qiao (PhD), and Brendan Jeon (PhD).
Graduated PhD students: Azam Asanjarani,
Ebby Thomas,
Tuan Dinh.
Further student supervision (at various times and of various sorts): Alexander Hodges, Angelique Husson, Andrew Liang, Beau White, Stijn Fleuren, Nikki Leijnse, Brendan Patch,
Aapeli Vuorinen, Darcy Bermingham, David Campbell, Kyle Young, Jiahao Diao, Julia Kuhn, Jori Selen, Kay Peeters, Liron Ravner, Manasa Anantharaman, Riley Vanderbyl, Tung Le, Wei Dong, Weichung Lai, Yifan Jiang, Yunqian Lin, Jiesen Wang, Matthew Richards, Thomas Graham, Wala Draidi, Alex Yan, Greg Marshall, Yang Leng, Yishan Peng, Yao Chen, Xuerong Wang, Rijul Jain, Leon Bitolkoski, (Julien) Minh Tram Tran, Jingye Liu, Hui Dong, Hoorain Malik, Pallavi Goswami, Taotao Pan, Hugh Roberts, Rick Heuijerjans, Eric Orjebin
[More...]
Future students: I am always looking for motivated students interested in pursuing a PhD. If you find the research themes above interesting,
contact me.
Major Additional Activities
At the moment, in addition to research and teaching, I am also engaged in the following:
Some of my notable past activities include:

I secured internal funding (over $0.5M) for the AI4Pandemics project and team that was operational during 2021 and 2022. See also our YouTube seminar series.

I was the Data Science Lead at the School of Mathematics and Physics. Within this role, I oversaw our school's part of the Masters of Data Science Program as well as all other school activities related to the Data Science discipline.

Initiation and codirectorship of the 2019 INFORMSAPS conference, held in Brisbane Australia.

Industry liaison at the School of Mathematics and Physics where I cocreated industry internships for credit.

Mathematics Honours coordinator at the School of Mathematics and Physics.

Various additional initiatives at the School of Mathematics and Physics.

In addition to INFORMSAPS, initiation and contribution to multiple other academic and industry, seminars, conferences, and workshops. [More...]

Director and creator of One on Epsilon  and codeveloper of the apps Epsilon Stream and Square Root Marbles.

Five years industry experience as a software and algorithm developer, team leader, and systems engineer  Israel aerospace industry.
Views
An academic webpage is not complete without a few personal items and some personal views.
I do quite a bit of (car) driving these years, mostly involving my kids, my wife Miriam and our dog Dash. So during this time, I get a chance to reflect and refine my views. There are a few opinion blog posts that I have written. Here are some more views:

On Artificial General Intelligence (AGI): It doesn't exist yet. A few years ago I thought that there isn't even a clue on how to make it. But since the rise of LLMs I think that sheer scale might do the job. Let's see...

On Israel, Gaza, and all that: This is a big subject where I believe that I am quite informed and I have many subtle views. In a nutshell I think: (1) Israel is slowly loosing its moral ground and has probably past the point of no return. It is becoming and aggressive Apartheid state. (2) Palestinian culture is violent at its core and is based on victim mentality and not on doing constructive stuff. It almost looses its definition if Israel vanishes. Worse of all, Palestinians don't seem to be looking for constructive solutions for building lives but rather converge on murderous organizations like Hamas.

On Ukraine: You won't see me waiving the Ukrainian colors because I don't support nationalism. However, I think the Ukrainian people are fighting a defensive battle not just for their freedom, but also for the freedom of the rest of the world.

On eating meat: Our world would be a better place if we don't.

On Global Warming: Yes. But it should be rephrased global variance (or volatility increase).
[More...]