Adaobi Ugwuonah

I've built and released an English-to-Igbo education-domain dataset, fine-tuned a LoRA adapter on Llama-4-Scout-17B, and published the model, dataset documentation, and research abstract on Hugging Face.

At the African Computational Linguistics Summer School, I engineered an Igbo speech corpus from over one hour of recorded audio, processed it with SoX and Python tools, and produced corpus statistics and visualisation plots. I also benchmarked Whisper across Igbo and Yoruba using WER and CER, and annotated Igbo speech data in Label Studio.

My work combines computational linguistics, acoustic phonetics, corpus building, ASR evaluation, and linguistic analysis for low-resource African languages. As a native Igbo speaker with fluent English and Yoruba, I'm focused on improving computational tools that reflect African language structure and use.

Alongside my NLP work, I've supported ASR training pipelines as a US English transcriber and annotator at AON Data, and I bring more than eight years of English-language teaching and education experience from OCECS and Teach For Nigeria.