Lab and computational work, mostly on the random side of cell biology. I am still learning most of this as I go.
University of Oxford
Working on how a molecule finds its binding site inside a cell that is already about thirty percent full of other molecules. The crowding slows diffusion down, but it also keeps the search inside a smaller space, so it is not obvious which of the two wins. I am trying to find the conditions where the second one does.
Laidlaw Foundation
A Laidlaw grant to spend a year on the queueing dynamics of mRNA transcription. Polymerases loading onto a gene get in each other's way, so the bursts that come out are not Poisson. I am fitting stochastic models to single-molecule data to see whether the non-monotonic activation they predict is actually there.
eBrain Lab, NYU Abu Dhabi
Fine-tuning pretrained language models on quantum chemistry, and building pipelines that predict molecular states before anything is synthesised. Most of my time goes on the cases where it gets things wrong, and on telling apart the ones that are the model from the ones that are the training data.
NYU Abu Dhabi
Computational tools and outreach for the university iGEM team, ahead of the 2026 conference in Paris. About half of it is modelling and half is explaining the modelling, to judges and to school students. The explaining is the harder half.
NYU Abu Dhabi
Worked with Professor Mazin Magzoub on how probabilistic RNA splicing is, writing code to characterise exon activity. The result stayed ambiguous under every analysis I tried. Reporting that, rather than going back for a cleaner one, was the part I found hardest.
University of Cambridge
A summer project on whether competition for resources can push soft matter to criticality, working from C. elegans oocytes and asking whether the cortical condensates are arranged hyperuniformly. Much of the summer went on rewriting the dry-lab pipeline instead, which was probably more use to the group than what I had gone there to do.
Institute of Molecular Genetics, Czech Academy of Sciences
Put together a Fiji and ilastik pipeline to plot phase portraits of protein dynamics during actomyosin cortex activation, with segmentation and track analysis across about fifty live-cell datasets.
NYU Abu Dhabi
Measuring how unstable cortical condensates are in C. elegans oocytes, from time-series microscopy. It came to something like a thousand trajectories, looking for transitions driven by noise rather than by anything in particular. This is where I started, and most of what I can actually do came out of it.