Independent researcher · Astana, Kazakhstan
Yerulan Mustafin
Mathematical logic, artificial intelligence, and financial modelling.
Model theory is the foundation of my mathematical work. I now use it to study how AI systems represent knowledge, support their answers, and handle conflicting sources. These questions also matter in finance, where conclusions depend on the data and the assumptions behind them.
01
Mathematics and artificial intelligence
My current project develops a model-theoretic framework for studying AI systems that use retrieved documents and knowledge bases to generate answers (RAG). Its central question is how a formal model of knowledge relates to the conclusions of an operational system.
I study the distinction between an answer being logically consistent and being supported by facts, and the conditions under which information from different sources can be combined consistently. This work connects my earlier research on Jonsson theories with current questions about knowledge representation and AI reliability.
A particular application I am interested in is financial research, where data change over time and analytical conclusions need to be checked against their sources.
02
Selected publications
A short selection. Titles are given as published, with links to journal records.
-
2002
Quelques propriétés des théories de Jonsson
The Journal of Symbolic Logic 67(2), 528–536.
-
2004
Structure des groupes linéaires définissables dans un corps de rang de Morley fini
Journal of Algebra 281(2), 753–773.
-
2006
Sous-groupes superstables de SL₂(K) et de PSL₂(K)
With Bruno Poizat. Journal of Algebra 297(1), 155–167.
-
2019
Constructing a Yield Curve in a Market with Low Liquidity
With S. Khakimzhanov, O. Kubenbayev, and D. Atabek. Russian Journal of Money and Finance 78(4), 71–98.
03
Professional practice
I lead TG Global, a sister company of Tansar Capital within Tansar Group. My work covers investment, financial risk, and research.
I headed the Risk Management and Strategy Department at the National Investment Corporation of the National Bank of Kazakhstan (NIC) in 2016–2017. From 2017 to 2022, I worked on financial stability and research at the National Bank of Kazakhstan.
My experience in Montreal includes financial analysis, trading strategies, and risk management at TitleTrading Services (2007–2011).
04
Academic background
In 1996–2000, I worked in the Laboratory of Logic and Computer Science at the Institute of Applied Mathematics in Karaganda on neural network models, classification, and decision-making under uncertainty.
I completed my doctorate in mathematics at Université Claude Bernard Lyon 1 in 2003 under the supervision of Bruno Poizat.
In Canada, I held a postdoctoral position and taught at Université du Québec à Montréal (2003–2004). In France, I taught and conducted research at Université Claude Bernard Lyon 1 (2004–2007).
At McGill University’s Department of Economics (2006–2007), I worked as a research assistant on economic models of central banking and the effects of central bank policy.
05