Dr. Vivian Oloo

Dr. Vivian Oloo is a Senior Researcher and NLP specialist at MCAAI, with a primary research focus on corpus creation and language model development for Dholuo — a Nilotic language spoken by over four million people in Western Kenya and a core target language for MCAAI's language technology programme.

Dr. Oloo leads the DhoNam dataset initiative, overseeing the end to end process of speech data collection, transcription, quality assurance, and open release. Under her leadership, the DhoNam corpus has grown to encompass 51.34 hours of native speaker audio from 59 speakers, carefully annotated across thematic domains including agriculture, healthcare, education, and conversational Dholuo. Her meticulous approach to corpus design has made DhoNam one of the most rigorously documented speech datasets produced by any African research institution.

Dr. Oloo brings deep expertise in transformer based neural architectures and their adaptation for morphologically complex languages — a significant technical challenge for Dholuo, which features agglutinative morphology and tonal distinctions that are difficult to capture using models pre trained primarily on Indo European languages. Her research addresses these challenges through targeted pre training strategies and phonological data augmentation techniques.

Alongside her research contributions, Dr. Oloo is a dedicated mentor who actively guides junior researchers and postgraduate students in corpus linguistics methodology, experimental design, and academic writing. She is widely regarded within MCAAI as one of the centre's most important capacity building figures.