28 August 2025
I build web interfaces for data input and exploration, and help researchers see their data through the eyes of researchers from adjacent fields and the public. I collaborate with scientists to transform their datasets into accessible interfaces, essentially translating between data and human sensibilities.
I'm designing and developing a data portal for the Amsterdam Time Machine that aggregates heritage data from Delpher (National Library of the Netherlands) and the Amsterdam City Archives. Unlike other time machines that use data to tell stories about the respective city, we tell a story about the data itself by visualising geo-located datasets across time and space through interactive heatmaps. This spatiotemporal indexing lets users explore without knowing what they're searching for and hopefully develop new research questions, or broaden the existing ones. We're launching with Amsterdam's 750th anniversary celebrations.
The DSC Away Days are great! As someone with a design background working among researchers, being part of the DSC helps me extend my thinking beyond information design. These collaborations clarified my role as a designer and developer since the DSC members approach data from different angles than what I'm used to.
My role has now evolved from simply utilising data to tell stories from my own angle, to opening the research and datasets so that others can use it to tell their own stories or come to their own conclusions.
Unsupervised machine learning methods, particularly dimensionality reduction and clustering. I've used Latent Dirichlet Allocation (LDA) to classify and extract HTML components from websites, and I'm currently using BERT's perplexity and other uncertainty metrics to detect creative language use in web scrape corpora. It's computers exploring without prior knowledge about the subject, so it aligns well with my work on explorative interfaces where people do more or less the same.
I've started using Typescript for the increasingly complex web apps I build at UvA, and Python for hobby machine learning projects. I used to obsess about performance too much, I had a Rust phase but realized low-level languages slow down my prototyping. Nowadays I think of myself as a Lego kit basher - I don't design the pieces, I assemble them. The engineers building the tools make this approach possible.