Research statement

I work across NLP and speech in roughly equal measure. On the language side: large language models and how they are adapted to new domains and languages, multilingual representation, and clinical and biomedical text. On the speech side: representation learning, prosody, recognition, and spoofing and deepfake detection. Multimodal models sit between the two and have taken more of my attention lately.

Evaluation runs through a lot of it — construct validity, contamination, whether a benchmark measures the thing it names. That has become a research area in its own right rather than a methodological footnote, particularly in Europe, where the AI Act is turning validity into something that has to be demonstrated rather than assumed.

These interests are broad on purpose. I am looking for a funded position on a defined project, and I would rather contribute to a group’s existing direction than arrive with a thesis of my own to defend. Most speech and language projects I read have something in them I would be glad to spend four years on, and I would rather be useful early than spend the first year negotiating scope.

I can work independently and I finish things: my studies have gone from question through to released code without day-to-day supervision, each reproducible from a single command. I am a working software engineer as well as a researcher, so experiments get built quickly and results get packaged so that someone else can run them, and I am comfortable enough below the research stack to set up, profile and instrument a pipeline myself rather than wait for someone to do it. I am also careful with results — I build the uninformative baseline before the model, state limitations plainly, and would rather report a well-controlled negative result than a positive one I cannot defend.

I am looking for a funded PhD position — a salaried research contract or a fully funded scholarship — in Europe or elsewhere. I am willing to relocate, I work in English, and I can start at short notice. I am happy to be assessed on a technical task, a code review of anything on this site, or a conversation about a project you already have funding for.