Under-resourced languages
Bringing modern NLP to languages with little digital text — corpora, models, and benchmarks for African and Indigenous languages.
About
I'm a Google DeepMind Academic Fellow at University College London. Previously I was a postdoctoral researcher at MBZUAI in the UAE, working with Prof. Thamar Solorio. I hold a PhD in Computer Science from Instituto Politécnico Nacional, Mexico, where I was advised by Prof. Alexander Gelbukh and Prof. Olga Kolesnikova.
My research focuses on natural language processing for the world's under-resourced languages — building multilingual language models, evaluation benchmarks, and speech and multimodal systems that serve communities typically left out of mainstream NLP.
Research interests
Multilingual LMs
Training and evaluating small and large multilingual language models that work across high- and low-resource languages.
Evaluation benchmarks
Culturally-aware, linguistically-honest benchmarks — so models are measured on what they actually need to do, not on convenient proxies.
Speech & multimodal
Speech recognition for African accents and clinical domains; vision-and-language datasets that reflect cultures outside the Western web.
Spotlight
The Zeno's Paradox of "Low-Resource" Languages
A critique of the "low-resource" label itself — arguing that the term collapses meaningfully different language situations and obscures what actually needs fixing.
MasakhaNEWS: News Topic Classification for African Languages
A benchmark for news topic classification across 16 African languages, with strong baselines and a public dataset.
See all publications →News
- 2026 2 papers accepted at ACL 2026: AfriMCQA-Multimodal Cultural Question Answering for African Languages and CommonLID: Re-evaluating State-of-the-Art Language Identification Performance on Web Data.
- 2026 Joined UCL as a Google DeepMind Academic Fellow.
- 2025 Joined MBZUAI as a postdoctoral researcher with Prof. Thamar Solorio.
- 2025 1 paper accepted at NAACL 2025: ProverbEval — LLM evaluation for low-resource languages.
- 2024 1 paper at NeurIPS 2024 D&B: CVQA — culturally-diverse multilingual VQA benchmark.
- 2024 2 papers at EMNLP 2024: Zeno's Paradox of "Low-Resource" Languages ★ Outstanding Paper & Walia-LLM (Amharic).
- 2024 1 paper at LREC-COLING 2024: EthioLLM.
- 2023 2 papers at EMNLP 2023; 1 at TACL (AfriSpeech-200); Best Paper at AACL (MasakhaNEWS); 1 at INTERSPEECH (AfriNames).