CV

Education

• Ph.D in Computer Science, University of Oxford, 2024 (expected)
• MSc in Statistics, University of Warwick, 2018
• BSc in Psychlogy, University of Groningen, 2017

Work experience

• Wise Payements (February to June 2022)

  • Core developer of Neural-Lifetimes. This library enables customer transactions prediction for various downstream applications using PyTorch.
  • Developed and implemented neural network model on unstructured time-series data.
  • Implemented SOTA methods for representation learning based on mutual information. Adapted methods that were only verified on MNIST to real-life data.

• 2021: Oxford Strategy Group Digital

  • Senior Consultant
  • Developed Machine Learning for financial industry client.

Skills

• Machine Learning
• Deep Learning
• Statistics
  • Decision Theory
  • Applied Frequentist and Bayesian Modelling
  • Time Series
• Bayesian Inference
  • Monte Carlo Simulation
  • Variational Methods
  • Bayesian Systems and Statistical Control
• Computer Vision

<!– Publications

Difference or delay? A comparison of Bayley-III Cognition item scores of young children with and without developmental disabilities

E-ViL: A Dataset and Benchmark for Natural Language Explanations in Vision-Language Tasks

Explaining Chest X-ray Pathologies in Natural Language

Certified Calibration: Bounding Worst-Case Calibration under Adversarial Attacks

A Stable, Fast, and Fully Automatic Learning Algorithm for Predictive Coding Networks

Fool Me Once? Contrasting Textual and Visual Explanations in a Clinical Decision-Support Setting

Towards Certification of Uncertainty Calibration under Adversarial Attacks

Benchmarking Predictive Coding Networks – Made Simple

Shh, don’t say that! Domain Certification in LLMs

–>

Teaching

Applied Machine Learning

Machine Learning

Reviewer

• ICLR 2025 Reviewer

Repositories

• FleissKappa a torchmetrics compatible implementation of FleissKappa.
• FilenameManager a very light-weight Python library that uses configuration parameters to create filenames and can convert filenames to parameters. Useful for saving ML experiment data to disk without database.
• LiteralCopy a simple Python library that takes existing objects in memory and returns a string with the syntax to recreate that object. Supports base Python, NumPy and PyTorch.

Service and leadership

• Created and maintained compute documentation for our research group
• Mentor at Project Access
• Honorary sailing instructor
• Chair of student representation during my undergraduate degree
• Winner of a 2016 transatlantic sailing race on the 52-foot yacht “Bank von Bremen”