I study how NLP systems can be made genuinely useful across different languages, audiences, and real-world contexts. This means building better evaluation frameworks, developing methods for content adaptation, and creating tools for under-resourced languages, particularly Welsh. Below are the three main themes that structure my current work.