Marie Curie PhD on Machine Learning for High Energy Physics, with a background in Theoretical Physics from Oxford

A significant portion of my research work is, in fact, highly relevant to the field of quantitative research. Below are some illustrative examples.

What I Work On

My recent focus has been on techniques directly applicable to quant finance, including:

  • High-frequency time series modeling
  • Low-latency inference on GPUs and FPGAs
  • Anomaly detection in high-dimensional streaming data
  • Systematic strategy design and backtesting
  • Stochastic processes and statistical inference

I’m particularly interested in applying mathematical modeling and machine learning to financial markets, in order to uncover structure in volatility regimes, alpha signal development, and market microstructure.

Selected Projects

Systematic Bitcoin Strategy Backtester

A trend-following backtesting framework written in Python with Pandas and NumPy. Computes performance metrics including:

  • PnL
  • Sharpe Ratio
  • Max Drawdown

Deployed a Freqtrade crypto trading bot on AWS with Docker, running 24/7 on a live Kraken portfolio.

Resources and Code: Project, Blog

Anomaly Detection on High-Frequency Data

Used a Spatio-Temporal Generative Adversarial Network (GAN) to detect rare events in real-time streaming data from 125+ sources. This method combines:

  • Graph Neural Networks (GNNs) to capture spatial features
  • LSTMs to capture temporal features
  • GANs to efficiently generate realistic time series matching the training data

The methodology applies naturally to:

  • Market anomaly detection
  • Regime shift identification
  • Volatility clustering

Resources and Code: Project

GNN-Based Real-Time Inference

Built a Graph Neural Network pipeline for fast particle tracking at CERN, achieving:

  • 100× GPU inference speedup (vs baseline)
  • 3× energy efficiency (on FPGA)

These optimizations are directly transferable to:

  • Latency-sensitive trading systems
  • Order book modeling
  • Market microstructure feature extraction

Resources and Code: Project

Blog Posts

You can find my blog posts related to finance and investing here.

Publications & Talks

You can find a full list of my academic work here, including several talks on fast ML for inference and detection—topics that align closely with low-latency trading and online portfolio monitoring.

Technical Stack

Languages: Python, C++, Julia, SQL
ML/Stats: PyTorch, TensorFlow, Scikit-learn, Keras, Statsmodels
Finance: Backtesting, alpha modeling, systematic strategy research
HPC/Deployment: CUDA, ONNX Runtime, Vivado HLS, Docker, AWS