A cohort-based research fellowship that pairs aspiring African NLP researchers with experienced mentors from leading labs and universities. Choose one of two tracks — build a dataset for your language, or pursue your own research idea — and get the guidance, mentorship, and GPU compute to see it through to a paper.
African languages are spoken by hundreds of millions of people and represented in almost none of the tools, datasets, and papers that shape modern NLP. The talent to change that already exists across the continent. What is missing, too often, is access: a mentor who has published before, who can sharpen a research question, read a draft honestly, and open the door to a first paper.
One thing moves the needle more than anything else — pairing motivated people with researchers willing to mentor them. The AfriNLP Fellowship is built on that idea: it connects Africa’s most promising NLP minds with experienced mentors from around the world, and makes sure the best guidance reaches people who would never otherwise have it, whatever their background.
Pick the track that fits where you are. Together with the AfriPlaybook — the guide to building African-language datasets — and AfriAnnotate, the tool to annotate them, the fellowship completes a full pipeline for African-language data: from learning how, to building it, to publishing it.
Get end-to-end guidance on creating a high-quality dataset for your language and a specific task — from scoping and annotation design to quality control and a documented release.
Bring your own idea — or take one from your mentor — and work on an open problem in African NLP. Not limited to data: modelling, evaluation, multilinguality, speech, and more.
You are paired with a mentor from a leading lab or university who meets you regularly, reviews your work, and guides the research from question to paper.
You work on an open problem in African NLP — your own idea or one matched to your interests — and own the work end to end with your mentor’s support.
The program is built around a concrete output: a paper submitted to a *ACL venue or the AfricaNLP workshop, with your mentor as a co-author and guide.
You join a cohort of peers across the continent — shared reading groups, work-in-progress sessions, and a network that outlasts the program.
No institutional affiliation, no big-name reference, and no prior publications required. We select on curiosity and effort, not pedigree.
Fellows get GPU compute for training and experiments, the AfriPlaybook’s templates and datasets, and the AfriAnnotate annotation tool — the whole pipeline in one place.
For fellows who want it, mentors offer optional guidance on applying to MSc and PhD programs — shaping a research statement, choosing labs, and navigating the process. Never required.
The first AfriNLP Fellowship cohort runs over roughly nine months, built around a single goal: getting each fellow’s work to a place where it can be submitted, reviewed, and published.
A short written application: who you are, what you want to work on, and why. No publications or referrals needed.
We review applications and pair each fellow with a mentor whose expertise fits their language, modality, and research interest.
A focused five-month program: weekly mentor meetings, monthly cohort check-ins, and steady progress from question to results.
Fellows write up their work and submit to a *ACL venue or the AfricaNLP workshop, with mentors guiding framing, reviews and rebuttals.
Every fellow is paired with an experienced researcher who guides them through the project — refining the question, reviewing experiments, and giving honest feedback on the writing. Mentors are drawn from leading labs, universities, and the African NLP community, and are matched to each fellow by language, modality, and research interest.
The aim is not a certificate. It is a real publication: a paper submitted to a *ACL venue (ACL, EMNLP, NAACL, or Findings) or the AfricaNLP workshop, co-authored with the mentor who helped build it.
Apply to join the first cohort, or step up as a mentor and help build the next generation of African NLP researchers. No affiliation, no referral — just the curiosity and the time to commit.
Students, self-taught researchers, and practitioners anywhere in Africa or the diaspora who work on — or are eager to work on — an African language or a problem that matters to African communities.
No. There is no requirement for an institutional affiliation, a referral, or a track record. We select on curiosity and effort, not pedigree.
There is no fee to take part in the fellowship. You only need the time to commit to the program.
Plan for a steady, part-time commitment across the roughly five-month research sprint — weekly meetings with your mentor and consistent progress between them.
Yes. The fellowship runs fully online, so you can take part from anywhere in Africa or the diaspora.
Yes. You pick the Data Track or the Research Track when you apply, and your mentor is matched to that choice. If you are unsure, tell us in your application and we will help you decide.
Research Track fellows are provided with GPU compute for training models and running experiments, so access to hardware is never the blocker.
Applications for the first cohort open in August 2026. Selection and mentor matching follow in October, with the program running through to a paper submission in 2027.
The fellowship grows out of — and connects fellows to — the communities already building African NLP.
