Research focuses on machine learning for scientific applications, especially in biomedical sciences
Integrates domain expert knowledge into predictive modeling and advocates for MLOps best practices
Works on mechanistic interpretability of transformer-based neural machine translation models, with emphasis on gender biases and non-binary gender identities
Uses Bayesian statistics for rigorous model evaluation in all technical work
Critically examines socio-technical implications of AI through phenomenology and participatory design, prioritizing inclusivity and ethics