About
Christina reckons with the blind spots in AI development - the languages models can't speak well, the labour they're built on, and the relationships they reshape. Her MSc (University of Edinburgh) achieved a 61% improvement in Tibetan-English machine translation by circumventing assumptions baked into dominant-language data and processing methods. Her research extended this to historical text recognition. With a background in UN labour migration and anthropology, she’s developing a social enterprise exploring how platform dynamics and algorithmic influence shape gendered violence and recovery.
Independent / otherEarly career / PhD (0-5 years)Computer scienceSocial sciencesOther