Sevi is News Labs’ Language Technology editorial lead. She develops, calibrates and employs tailored language technology solutions for BBC World Service to facilitate multilingual production in 43 languages.
Her current focus is building custom MT models for 16 under-resourced languages in the EU-backed GoURMET project with some of Europe’s leading MT researchers. Trained on genuine newsroom content, the models are built into production prototypes improving efficiency and value while breaking down language barriers across hubs. A recent example of such prototypes aggregates and filters content and uses recommendation algorithms to suggest machine translated content in bilingual or trilingual form to individual journalists in search of sustaining content.
Previously a journalist, translator, interpreter and project manager, Sevi is also working with journalists and researchers to help multimedia journalists embrace transcription, digital editing, TTS & STT tools in their workflows.