
Independent company · Johannesburg, South Africa
Lelapa AI
South African language AI company building resource-efficient models for African languages, maker of Vulavula and the open InkubaLM.
Johannesburg, South Africa
- Founded
- 2022
- Type
- Independent company
- Based in
- Johannesburg, South Africa
- Key papers
- 1
- Featured projects
- 7
About
Lelapa AI was founded in Johannesburg in 2022 by Pelonomi Moiloa, Jade Abbott and academic collaborators, and the name means home in Sotho and Tswana. Moiloa is chief executive and Abbott is chief technology officer, and the team page also lists Vukosi Marivate as chief information officer, Benjamin Rosman as chief scientist and George Konidaris as an advisor. Brown University and Wired have described Moiloa, Abbott, Marivate, Konidaris, Rosman and Pravesh Ranchod as the founding group.
The company's commercial product line is Vulavula, with Vulavula Transcribe for multilingual contact centres and Vulavula Translate for translation between African languages. It targets telecoms and financial services.
Lelapa also releases research models and datasets openly, notably InkubaLM, and states that it builds resource-efficient language AI designed to scale.
Mission and ambitionsLelapa states that it builds resource-efficient language AI for global scale, starting with African languages that are poorly served by existing models.
Known for
- 01
Vulavula speech and translation products
- 02
InkubaLM small African-language model
- 03
Founded by Masakhane and academic researchers
Milestones
- 2022
Lelapa AI founded
Founded in Johannesburg to build language AI for African languages.
- 2022
Vulavula
Launched speech and translation services for African languages.
- 2024
InkubaLM
Released a small open multilingual model for African languages.
- 2025
Vulavula Transcribe and Translate
Current product line aimed at contact centres and translation between local languages.
Open source
Notebook for the Inkuba challenge.
Vulavula examples.
Translation demo.
- InkubaLM-0.4Bmodel
InkubaLM.
- Inkuba-instructdataset
Instruction data.
- Inkuba-Monodataset
Monolingual data.
isiXhosa speech.
Research papers
Notable releases
Openness
Research models and datasets are on Hugging Face (InkubaLM-0.4B is listed under CC BY-NC 4.0); the commercial Vulavula product is closed.
People
- Pelonomi MoiloaChief Executive Officer
- Jade AbbottChief Technology Officer
- Vukosi MarivateChief Information Officer
- Benjamin RosmanChief Scientist
- George KonidarisAdvisor
Listed from the lab's own pages and public profiles.
Contact and community
Contact details and links are on the lab's website (https://lelapa.ai); code questions go through issues on their GitHub organisation.
Sources
Compiled from the lab's public pages, Hugging Face and GitHub. Last checked 2026-11-17. Something wrong or out of date? Tell us on Discord.



