Public projects
and experiments.

A community-driven directory for digital sovereignty: find GDPR-compliant, European-owned alternatives to Big Tech. The project also includes a community hub, blog and contribution path.

A browser-based AI content disclosure tool. Users can add labels to images, video, audio, PDFs and ZIP files while the file stays local in the browser.

Investigates whether human similarity judgments can improve machine-vision embeddings by fine-tuning ResNet-50 models on THINGS object images and human odd-one-out data.

An end-to-end research pipeline for collecting and analysing publicly shared ChatGPT conversations from Reddit, including semantic, sentiment, linguistic and syntactic alignment.

A card-deck interface for the Data Is Plural archive. Swipe through datasets and narrow the field toward ten datasets you are most likely to enjoy.

A retrieval-augmented generation project for exploring meeting transcripts, summaries and project data.

A web application that generates flashcards from YouTube video transcripts, turning educational content into material that is easier to study and remember.

An interactive workshop and published booklet for incoming MSc Cognitive Science students, covering Python refreshers, notebooks, data work and model workflows.