r.michael(at)posteo.net | github | scholar | CV | bsky
I'm a Ph.D. fellow in Computer Science (Machine Learning) at the University of Copenhagen (DIKU), supervised by Wouter Boomsma at the KU BioML group. I develop probabilistic models and optimization methods with applications in the experimental design of proteins and small molecules. My previous work includes supervised learning for protein function prediction with representation assessesments for engineering [MKH24], and high-dimensional Bayesian optimization for discrete inputs [GD24,M24]. Currently, I investigate active learning for out-of-distribution generalization related to protein function landscapes. I regularly collaborate with biochemists [MKH24,MD25], as well as StatML researchers [MBS25], and develop research software running on HPC at scale [GD24]. My background is in Bioinformatics [RM21] with a prior in Cognitive Science. I'm a DDSA PhD grant holder for "Principled Bayesian Optimization and scientific ML for Protein Engineering", my supervisor's group is part of the Center for Basic Machine Learning Research in Life Science (MLLS), and I'm also a member of the Pioneer Centre for AI, first PhD cohort.
Status: Running Bayesian AL experiments on refined task formulations with collaborators and guidance from the RainML group, OxCSML.
Status: Computing Ds-optimal ligand libraries from dose-response curves toward three target enzymes for iterative lead-refinement. Collaboration with the Hatzakis Lab, CHEM at the University of Copenhagen.
Status: Initialized collaboration with Junghyun Lee at OptiML Lab, KAIST.
Stack: trieste, tensorflow, mlflow
Stacks: torch, ax, wandb
(*) indicates equal contribution
Code & Preprint ▼Stack: torch, pyro, biopython, matlab