The Collective Village

Prof. Vukosi Marivate, the long patience of building for the future, and the village that keeps growing

Masakhane means we build together. It is a good name, and it is a promise. Some of the people who first spoke it into being are still here, still building, seven years on.

This feature is about one of them, Prof. Vukosi Marivate, a Masakhane community member, and it is truly a privilege to write it.

Start where it started: with a theme.

In 2018, the Deep Learning Indaba, (might want to give a bit of info about it for those who might not be familiar?) which Vukosi co-founded, took Masakhane as its theme. Jade Abbott and Laura Martinus heard it as a call and established the beginnings of the Maskhane community. By 2019, when Indaba was held in Kenya, the call had become not only a community, but an open, continent-wide, distributed effort to build machine translation for African languages, by Africans, together.

Vukosi was part of the team that was there from the very start. Seven years later, he continues to make room so more of us can be a part of this monumental effort.

EVERYONE, FIRST

One of the community’s first major papers says a great deal about the kind of builder Vukosi is.

Participatory Research for Low-resourced Machine Translation: A Case Study in African Languages (Findings of EMNLP 2020) produced new translation datasets and machine translation benchmarks for more than 30 African languages, with human evaluation for a third of them. It showed that people with no formal training in NLP could make a genuine scientific contribution when the research is designed to let them in. It went on to win the inaugural Wikimedia Foundation Research Award in 2020.

Vukosi’s name is on it, with the University of Pretoria beside it, among dozens of others. But in Masakhane’s foundational papers the first author is not a person at all. It is a symbol, ∀, for all, standing in for the whole community, with everyone else’s order randomised in quiet rebellion against the status that author order confers.

That is a choice made by people who already had names worth putting first and chose to put the community there instead.

The first name on the paper is everyone’s.

THE COST OF WAITING

Vukosi has a memorable way of describing what it costs to leave a language out of technology: it is like a high-interest credit card. The longer the debt sits, the more expensive it becomes to settle.

So he does not let it sit.

At the University of Pretoria, he founded and leads the Data Science for Social Impact (DSFSI) research group, which uses local challenges as the springboard for research: science, energy, public safety, utilities, public health. When the pandemic arrived, DSFSI built the COVID-19 Data Repository for South Africa (COVID19ZA), an open public data infrastructure for which he was principal investigator.

For languages, the lab’s habit is the same: develop the resource that is missing, then give it away. Preparing the Vuk’uzenzele and ZA-gov-multilingual South African multilingual corpora gathered government speeches and the government newspaper Vuk’uzenzele, translated into all 11 official languages of South Africa, into corpora anyone can build on. TextAugment, the lab’s open library for stretching small text datasets further, won the Human Language Technology Research Software award at the 2026 NSTF-South32 Awards.

The newspaper’s name means, roughly, wake up and do it yourself. It became the lab’s motto, with one addition.

Do it yourself. Then do it for everyone.

WE BUILD TOGETHER

Vukosi’s path starts in Ga-Rankuwa, a township north of Pretoria, and keeps finding its way home.

A BSc and an MSc in Electrical Engineering from the University of the Witwatersrand, the master’s under Prof. Tshilidzi Marwala. A PhD in Computer Science at Rutgers University as a Fulbright Science and Technology Fellow, supervised by Prof. Michael Littman, along the way  in 2012, a Google internship in Mountain View. Then back home: the CSIR, Council for Scientific and Industrial Research. South Africa’s national research and development organisation, based in Pretoria and then the University of Pretoria, where he is a Professor of Computer Science and holds the ABSA UP Chair of Data Science.

Read in continuous prose, it is a distinguished career. Internalised carefully, this is the story of a person who keeps coming home and creating room for other people. The Deep Learning Indaba. Masakhane. Lelapa AI, built on AI for Africans, by Africans. And the African Institute for Data Science and Artificial Intelligence (AfriDSAI), which he directs, backed in 2025 by a US$1 million gift from Google for MSc, PhD and postdoctoral fellowships. He has spoken about watching young people join his classes and research groups since 2006 and grow into experts in their own right. That, more than any title, is the through-line.

Inside our community, his name sits on work that has shaped and articulates how we think. The Esethu Framework (ACL 2025), which readers of our last feature will remember, rethinks how datasets in low-resource languages are governed so that the communities who give their language keep a real stake in what results from it. Its proof of concept, the Vuk’uzenzele isiXhosa Speech Dataset, grew from the same South African ground.

With African Next Voices, he led the South African arm of what is thought to be the largest dataset of African languages for AI yet: 9,000 hours of speech in 18 languages across Kenya, Nigeria and South Africa, recorded in everyday settings like farming, health and education. The South African dataset, Swivuriso, holds around 3,000 hours in seven languages: isiZulu, isiXhosa, Sesotho, Setswana, Xitsonga, Tshivenda and isiNdebele. Its terms forbid anyone from using those voices for voice cloning or speech synthesis. Consent, written into the data itself.

Asked why any of this matters, he answered in five words: “Language is access to imagination.”

A room is only a room until someone holds the door open. Then it becomes a village.

A YEAR OF RECOGNITION

In February 2026, the UN Secretary-General recommended Vukosi for the United Nations Independent International Scientific Panel on Artificial Intelligence. He is one of 40 experts chosen from more than 2,600 applications from over 140 countries, carrying the experience of African language technology into the room where the world takes stock of AI.

Then, on 19 May 2026, at the Sefako Makgatho Presidential Guesthouse in Pretoria, President Cyril Ramaphosa conferred on him the Order of Mapungubwe in Silver, one of South Africa’s highest national honours. The citation recognised his contributions to data science, artificial intelligence and natural language processing, and how they have advanced technological capability across South Africa and the continent. An order named after one of Africa’s great ancient civilisations, given to Vukosi for helping ensure that African languages have a future in the world’s newest technology. It is hard to imagine a better fit.

The recognition arrived in good company: the 2025 SAICSIT Emerging Pioneer Award, and his team’s NSTF-South32 award for TextAugment.

And characteristically, when he spoke about the honour, he spoke about everyone else. He admitted he had wrestled with feeling worthy of it. Aptly, when he looked at the award, he said he saw a mirror reflecting “a vast, beautiful village of researchers, students and community-builders” who had chosen to build a shared vision with him. He called his own reflection a win for our collective village.

“Our languages, our data and our perspectives are inherently valuable.”

- Prof. Vukosi Marivate, on receiving the Order of Mapungubwe in Silver, 2026

A village. He named it with ease. Every person who has ever translated a sentence for a Masakhane benchmark, annotated an entity, reviewed a paper, or shown up on a community call lives there.

WHAT WE ARE BUILDING

In 2019, the idea that African languages would have open benchmarks, community-governed datasets, thousands of hours of recorded speech, their own research institutes and a voice on a UN scientific panel was a hope held by a small number of motivated, hardworking people in a handful of rooms.

Now: there is a Masakhane community with more than 4,000 members from 49 African countries and the diaspora. There is a Deep Learning Indaba that travels the continent. There is AfriDSAI. There is Swivuriso, and the Esethu license, and a generation of students who learnt the field from people who believed their languages belonged at its centre.

Africa currently produces about 0.9 percent of the world’s AI research, by Vukosi’s own count. He has spent his career changing that number. So have all of us.

That is what we have been doing all along. Not waiting for the world to arrive.

Building the village.

Prof. Vukosi Marivate is Professor of Computer Science and ABSA UP Chair of Data Science at the University of Pretoria, Director of AfriDSAI, a co-founder of the Deep Learning Indaba, Masakhane and Lelapa AI, a member of the UN Independent International Scientific Panel on AI, and a 2026 recipient of the Order of Mapungubwe in Silver. His work is at vima.co.za and dsfsi.github.io.

S O U R C E S

Every fact in this feature is drawn from the public record: university and government announcements, the ACL Anthology, arXiv and press coverage. The words attributed to Prof. Marivate are reproduced from the University of Pretoria’s and Bizcommunity’s coverage of the 2026 National Orders, and from BBC reporting on African Next Voices.

- Vukosi Marivate - personal site. https://www.vima.co.za

- Vukosi Marivate - Wikipedia. https://en.wikipedia.org/wiki/Vukosi_Marivate

- University of Pretoria, UP academics receive National Orders for reshaping South Africa’s future through AI and the arts (May 2026). https://www.up.ac.za/news/academics-receive-national-orders-reshaping-south-africas-future-through-ai-and-arts

- Bizcommunity, UP academics receive National Orders. https://www.bizcommunity.com/article/up-academics-receive-national-orders-for-reshaping-south-africas-future-390928a

- Wits University, Wits alumni shine at 2026 National Orders ceremony. https://www.wits.ac.za/news/sources/alumni-news/2026/wits-alumni-shine-at-2026-national-orders-ceremony.html

- AfriDSAI, Prof. Vukosi Marivate awarded the Order of Mapungubwe in Silver. https://www.up.ac.za/afridsai/news/celebrating-excellence-afridsai-director-Prof-vukosi-marivate-awarded-order-of-mapungubwe

- SAnews, Government congratulates Prof. Marivate (February 2026). https://www.sanews.gov.za/taxonomy/term/4?page=182

- University of Pretoria, Prof. Vukosi Marivate appointed to UN Independent International Scientific Panel on AI (via BusinessTech). https://businesstech.co.za/news/?p=853645

- University of Pretoria, Google donation boosts African-led AI research at UP. https://www.up.ac.za/department-of-institutional-advancement/news/google-donation-boosts-african-led-ai-research

- IT-Online, UP receives $1m Google donation for AfriDSAI. https://it-online.co.za/?p=301878

- Participatory Research for Low-resourced Machine Translation: A Case Study in African Languages (Findings of EMNLP 2020). https://aclanthology.org/2020.findings-emnlp.195/

- Masakhane — Machine Translation for Africa. https://arxiv.org/abs/2003.11529

- Slator, The Masakhane project puts Africa on the machine translation map. https://slator.com/the-masakhane-project-puts-africa-on-the-machine-translation-map/

- The Esethu Framework: Reimagining Sustainable Dataset Governance and Curation for Low-Resource Languages (ACL 2025). https://aclanthology.org/2025.acl-long.1487/

- Swivuriso: The South African Next Voices Multilingual Speech Dataset. https://arxiv.org/abs/2512.02201

- Africanews / BBC, The linguistic diversity of Africa and its absence in AI development. https://www.africanews.com/2025/09/05/the-linguistic-diversity-of-africa-and-its-absence-in-ai-development/

- TextAugment library (DSFSI). https://github.com/dsfsi/textaugment

- Data Science for Social Impact. https://dsfsi.github.io/

- Masakhane — a grassroots NLP community for Africa, by Africans. https://www.masakhane.io/home