Publications

Peer-reviewed work

Five papers on natural language processing, speech and computer vision, all of them concerned in one way or another with making machine learning work for languages and settings that the field has largely overlooked.

5 publications 2 arXiv preprints 2 Springer 1 OpenReview

2025

Sunflower: Expanding African Language Coverage in LLMs

arXiv

On extending large language model coverage to African languages that are poorly represented in existing training data and evaluation suites.

2025

Speech Data Requirements for ASR in African Languages

arXiv

How much speech data automatic speech recognition actually needs in low-resource African language settings, and what that implies for anyone planning a collection effort.

2024

Real-Time Spoken Instruction & Translation in Ugandan Languages

OpenReview

Real-time spoken instruction and translation for Ugandan languages — the research thread that underpins the speech-to-speech translation work.

2023

Explainable AI for Black Sigatoka Detection

Springer

Making crop-disease classification interpretable, so that a farmer can see the basis for a diagnosis rather than being asked to trust an opaque model.

2023

Transparent Intelligent Vision for Black Sigatoka Detection

Springer

Computer vision for detecting Black Sigatoka in banana plantations, with transparency of the model's reasoning treated as a design requirement.

Elsewhere