CV

Software Engineer (part-time)

ScreenPoint Medical

• Developed and maintained automated tests for large-scale medical AI software (2D/3D medical image processing).
• Automated manual verification steps and led a project to produce a shorter, more maintainable verification report.

AI Research Intern

ScreenPoint Medical

• Built a robust 3D MRI segmentation pipeline for breast density estimation, now used for internal R&D.
• Proposed a style-randomization augmentation for domain generalization across MRI sequences; first-author paper at MIDL 2024.

Data Science Intern

Arcadis

• Built and deployed machine learning dashboards for housing price analysis and for the optimal placement of EV parking spots and charging poles.
• Presented findings to stakeholders and advised on data-driven business decisions (Python, SQL).

PhD

Donders Institute, Radboud University

Artificial Intelligence, Machine Hearing

• Research: machine hearing, self-supervised (spatial) audio and speech representation learning, and spatial audio LLMs.
• AudioSphere (NeurIPS 2026): first sound event localization and detection from a frozen self-supervised backbone; 1.9° median direction-of-arrival error on a planar microphone array.
• SpeechJEPA beats wav2vec 2.0 with one hour of labels while using about 17× fewer training FLOPs.
• Bearings raises the location-aware SELD F-score from below 4 to 50 on TAU-NIGENS 2021.
• Supervisors: Kiki van der Heijden and Marcel van Gerven. Member of the ELLIS Unit Nijmegen.

MSc

University of Amsterdam

Artificial Intelligence

• GPA: 4.0/4.0.
• Thesis: Domain Adaptation of Dense Retriever Models.

BSc

Radboud University

Artificial Intelligence

• GPA: 3.96/4.0.
• Teaching assistant for Functional Programming and Bayesian Statistics.

AudioSphere: Towards Self-Supervised Spatial Audio Representation Models

NeurIPS 2026

University of Amsterdam at the CLEF 2024 SimpleText Track

Working Notes of CLEF 2024, CEUR Workshop Proceedings 3740

Style Randomization Improves the Robustness of Breast Density Estimation in MR Images

MIDL 2024 (PMLR 250)

Reproducibility Study of “Quantifying Societal Bias Amplification in Image Captioning”

NeurIPS 2023 Main Hall Poster, ML Reproducibility Challenge 2022

SpeechJEPA: Clustering-Free Latent Prediction for Compute-Efficient Speech Representation Learning

Under review at ICASSP 2027

Bearings: Self-Supervised Soundfield Embeddings from First-Order Ambisonics

Under review at ICASSP 2027

EquiSELD: Efficient Training of Equivariant Sound Event Localization and Detection Networks

Under review at ICASSP 2027

WavJEPA: Semantic Learning Unlocks Robust Audio Foundation Models for Raw Waveforms

Under review at IEEE/ACM TASLP

GRAM: Spatial General-Purpose Audio Representations

arXiv:2506.00934

Remining Hard Negatives for Generative Pseudo Labeled Domain Adaptation

arXiv:2501.14434

Interpretability Analysis of Domain Adapted Dense Retrievers

arXiv:2501.14459

On Correlating Factors for Domain Adaptation Performance

arXiv:2501.14466