AfirkaLLM Part 2: What teaches an AI to speak African languages?
Five controlled experiments, each with one change, judged against the untrained model. The answer wasn't what we expected.
Leading Research in Data Science for Social Impact
We bridge the gap between cutting-edge data science research and real-world societal challenges, with a special focus on African language technologies.
Our Impact Areas
Research that drives meaningful change in society
🌍 Data Science for Society
Developing methods and tools that enhance decision-making capabilities while ensuring user-centered approaches to data science research.
🗣️ African Language NLP
Advancing natural language processing for African languages through innovative tools, datasets, and methodologies.
🤖 Machine Learning Research
Developing ML approaches that address real-world challenges while considering ethical and societal implications.
Recent Research Highlights
Our latest contributions to the scientific community
Our research spans multiple domains, from developing language technologies for under-resourced African languages to creating data-driven solutions for societal challenges. We publish in top-tier venues and actively contribute to open science through our datasets, models, and tools.
Open Resources & Tools
We believe in open science and making our work accessible to all
Datasets
Curated datasets for African language research and social impact studies
Software & Models
Open-source tools and pre-trained models for researchers worldwide
Documentation
Comprehensive guides and tutorials for using our resources
Featured Projects
Masakhane Translate
Machine translation platform for African languages, making communication across language barriers more accessible.
Masakhane NLP
Grassroots organization focused on advancing NLP research and applications for African languages.
COVID-19 ZA Data
Comprehensive data repository tracking COVID-19 in South Africa, supporting research and policy decisions.
Latest News
AfirkaLLM Part 1: Africa is 0.057% of the web. Here is how we grew it.
Before you can teach a model a language, you need its text. Most of Africa's data was mislabelled, uncrawled, or never online.
Probabilistic Machine Learning with David Ríos Insua
Image: Participants of the “Probabilistic Machine Learning” short course on campus at the University of Pretoria.
DSFSI Research Earns Multiple Honours at ACL 2026
Image: Dr Idris Abdulmumin with the Social Impact Paper Award and SAC Highlight certificates for Afri-MCQA at ACL 2026 in San Diego.
Join Our Research Community
Interested in contributing to cutting-edge research that makes a difference? We welcome collaboration and new members.