About

Oxford PhD researcher in Generative AI for market microstructure. I study the realism of GenAI-generated order flow through the lens of market impact, combining tokenized Limit-Order-Book message streams with sequence models (SSMs, transformers, diffusion), GPU-accelerated JAX simulations, and reinforcement learning for quantitative trading. Earlier, I worked on portfolio optimization and causal inference. I’m a member of Oxford Engineering Department, my supervisors are Stefan Zohren and Jakob Foerster.

My interests center on state-of-the-art machine-learning methods for trading strategies, market making, execution, and predictive analytics — with additional work in causal inference, graph neural networks for time-series graphs, regime-switching models, and online portfolio selection.

Previously at Samsung R&D, I built ML for sensor time-series (signal processing, dependency structure modeling, end-to-end feature delivery). In academic research (Skoltech–Sberbank), I worked on uncertainty quantification, anomaly detection, and deep learning. Earlier, at a top-ranked regional bank, I built ATM withdrawal prediction and a discrete-optimization system for cash logistics.