Contribute

MULAMU is a collaborative, open project. Whether you are a clinician, linguist, engineer, or funder, there is a meaningful way for you to help advance maternal health AI for African languages.

Ways to contribute

Data collection

Health workers · Community members · Clinicians

Our models are only as good as their training data. We actively collect prenatal and peripartum Q&A pairs in Luganda, Runyankore-Rukiga, and Swahili with frontline health workers and clinical staff in Uganda.

  • Record or transcribe question-answer conversations in supported languages
  • Provide in-context cultural and clinical feedback on collected samples
  • Help identify gaps in coverage (conditions, regions, dialects)
Get in touch

Annotation & evaluation

Clinicians · NLP researchers · Bilingual speakers

High-quality annotation is the backbone of safe clinical AI. We need people who speak the target languages and understand maternal health to review and label model outputs.

  • Review model-generated answers for clinical accuracy and safety
  • Label and correct translations between English and target languages
  • Evaluate response fluency and cultural appropriateness
Get in touch

Engineering & code

ML engineers · Software developers · DevOps

From model fine-tuning pipelines to the audio inference stack, there is always engineering work to be done. All of our code and models are published openly on Hugging Face.

  • Improve model training, quantisation, and inference pipelines
  • Build evaluation harnesses and benchmark suites
  • Contribute to the demo interface and API layer
  • Help with data pre-processing and cleaning tooling
Get in touch

Research collaboration

NLP researchers · Global health academics · Clinicians

We welcome formal and informal research collaboration — whether that's co-authoring a paper, advising on clinical methodology, or providing compute or funding for experiments.

  • Joint evaluation of models on new benchmarks or populations
  • Clinical study design and protocol review
  • Co-authorship on papers and technical reports
Contact the team

Translation & localisation

Bilingual speakers · Linguists · Community translators

We are expanding to additional African languages. If you are a fluent speaker of a low-resource language spoken in a maternal health context, we want to hear from you.

  • Translate existing English Q&A pairs into new languages
  • Review and validate automated translations
  • Help document linguistic and dialectal variation
Get in touch

Funding & partnerships

Funders · NGOs · Academic institutions · Industry

Sustaining open research in low-resource settings requires dedicated support. We are open to grant partnerships, institutional collaborations, and compute donations.

  • Research grants and fellowships
  • Compute credits (cloud GPU / TPU)
  • Institutional research partnerships
  • In-kind support (clinical access, community networks)
Reach out

Our commitment

What we stand for

Three principles that guide every decision we make — from data collection to model deployment.

01

Open by default

Transparency

All models and datasets are published openly on Hugging Face under permissive licences. Contributors' work is credited and made freely available to anyone who needs it.

02

Community-first

Equity

Data is collected in partnership with local health workers. We do not extract value — we build capacity and share authorship with the communities closest to the problem.

03

Clinically responsible

Safety

Every contribution is reviewed for clinical safety. We do not deploy outputs without human oversight by qualified health professionals.

Start contributing today

Not sure where to begin? Send us a message and tell us a little about your background. We'll match you to the track where you can have the most impact.