About

I am a Principal in Portfolio Management, Quant Research and Technology at the European Bank for Reconstruction and Development (EBRD), working at the intersection of quantitative research, artificial intelligence and portfolio management. Alongside active portfolio management, I develop quantitative and AI-driven tools for investment decision-making, portfolio analytics and optimisation. I lead development of the Virtual Portfolio Manager (VPM), a major focus of my work in applied machine learning and agentic AI.

I am also completing an MSc in Machine Learning and Data Science at Imperial College London. My research interests span statistical machine learning, agentic AI and quantitative finance, with a particular focus on how data representation affects prediction. My recent research with Professor Niall Adams investigated representations of irregular event histories for supervised learning under incomplete follow-up.

I hold an MSc in Finance and Economics from the London School of Economics and a Master’s degree in Applied Mathematics and Computer Science from Lomonosov Moscow State University. I have also completed the Certificate in Quantitative Finance.

Virtual Portfolio Manager

The Virtual Portfolio Manager (VPM) is a major project I lead at the EBRD, bringing agentic AI into investment analysis and portfolio research. I work with colleagues across Treasury, AI and technology to connect portfolio and market data with quantitative tools for investment analysis and trade research, with portfolio managers retaining oversight. My contribution combines quantitative research, development of the agentic system and coordination across the teams building and evaluating it.

Agentic workflows

I build the agentic system that connects analytical workflows to approved quantitative tools and relevant data. The emphasis is on sourced, reproducible analysis that portfolio managers can review. This work brings together the capabilities of AI systems with the practical requirements of an investment desk: useful analytical outputs, reliable access to data and a clear relationship between the analysis and the information supporting it.

Machine learning and quantitative foundations

My approach to VPM draws on my experience developing quantitative methods and research infrastructure for Treasury. That work includes interest-rate models, forward-curve bootstrapping, mean–variance portfolio optimisation, and research into machine-learning and deep-learning trading signals. I also develop reusable data pipelines and analytical applications for quantitative research, portfolio monitoring and scenario analysis. This combination of financial modelling, machine learning and data engineering informs how I approach the project and the analytical tools it can bring together.

Evaluation and collaboration

Reliability, reproducibility and human oversight are central to the development process. I work with colleagues to translate portfolio-management needs into analytical workflows and prepare the system for production testing and internal review. The project also involves presenting quantitative research and agentic AI capabilities to senior leadership, connecting technical development with operational requirements and governance. For me, VPM is an opportunity to bring my experience as a portfolio manager and quantitative researcher into the development of practical AI systems, while keeping evaluation and the investment professional’s judgement central to their use.

Research

Event histories

My recent research with Niall Adams investigated statistical representations of irregular event histories for prediction under incomplete follow-up. The study used separate chronological cohorts for encoder estimation, model fitting and evaluation to control information leakage. My continuing interests include how these representations behave under censoring and across related prediction tasks.

Internal quant research

Since 2016, I have carried out internal quantitative research at the EBRD on portfolio risk models, macro signal generation, and regional asset-liability frameworks, for sovereign clients and for departments inside the bank.

Regime-switching allocation

With A. Chertok and N. Alexeychuk (2017), supervised by Viktor Korolev, I studied how to allocate limited resources when a system moves between unobserved regimes, using a Gumbel reparametrization. The paper was published in the collected papers of young scientists at the Faculty of Computational Mathematics and Cybernetics, Moscow State University.

Interbank liquidity

With Kathy Yuan at the London School of Economics, I worked on network risk in interbank liquidity. The paper, Network Risk and Key Players: A Structural Analysis of Interbank Liquidity (E. Denbee, C. Julliard, Y. Li, and K. Yuan, 2016), received a Fondation Banque de France research grant.

The market for news

Supervised by Vladimir Ulyanov, my graduation thesis at Moscow State University (2015) developed a stochastic model of the market for news. It was included in Theses of Best Graduation Works 2015, published by the Faculty of Computational Mathematics and Cybernetics.

Ventures

I founded NFT4U (2021–2022), a marketplace and studio for digital artists, and built the platform in Python, Solidity, and Django.

Before that I co-founded QuantSystem (2015–2016), a MATLAB system for backtesting financial strategies, shown at a MATLAB financial modelling conference.

News

  • 11/2026. I will speak at QuantMinds International in London, on Volatility and tail risk: Managing the extremes (16 November).