Designed and implemented deep-learning pipelines in PyTorch, training and evaluating models on large-scale data collected by the H.E.S.S. gamma-ray detector. Built analysis pipelines for time-series data from NASA's Fermi observatory.

Trained a transformer to segment Mars-analogue terrain from Atacama rover imagery. Built EKF localisation from IMU, rotary-encoder and visual-odometry data.

Exploring deep learning to improve bandwidth efficiency through link-history-based modulation and coding selection. Built a DVB-S2 MATLAB/Simulink setup with fixed-point and HDL workflows.

Built a statistical pipeline to identify spectral and temporal variability in NASA Fermi observations associated with anomalous events, leading to a first-author research paper.

Designed and built a solar tracker and spectrometer for in-flight data acquisition. Proposed the project, led the development team and secured competitive university funding.

Tutored undergraduate and postgraduate students in computer science and programming using Python and C++.

Applying ML and statistical methods to large-scale scientific data. Coursework: Advanced Deep Learning, Time-Series ML, ML in Communications, NLP and Artificial Intelligence.

Coursework in supervised learning, SVMs, PCA, CNNs, ResNet, vision transformers, detection, segmentation, RNNs, LSTMs, GNNs, multimodal learning, MLOps and edge deployment.

Eastern European Machine Learning Summer School, Cetinje, Montenegro, 27 July to 1 August 2026. I presented a poster on deep learning for particle detectors.

PUDL turns public US energy records into analysis-ready data.

CLIMADA is an open-source Python framework for climate risk assessment.