Skip to content
Somast
Company profile

Somast company profile

Somast is an open research lab and education company building Somali language data, models and training.

Download PDF (English) Download PDF (Somali)Last updated Oct 1, 2026

Who we are

Somali is spoken by more than 20 million people, yet it is almost absent from the data that trains modern AI. Somast collects and publishes open Somali text and speech, trains Somali models on it, and teaches people to build with it. The company was founded in 2023 and works from Mogadishu.

Our mission

Open AI infrastructure for the Somali language, built with the community.

Our values

  • Open by default

    Datasets, model weights, and evaluation code are released openly so anyone can build, verify, and improve them.

  • Community-owned

    The people who speak the language help build the technology, and own its direction.

  • Rigorous & reproducible

    Clear benchmarks, documented methods, and results others can reproduce and extend.

Numbers

Each figure has the date it was last checked.

20M+
people speak Somali

Estimate used across our materials

Oct 1, 2026

3
open datasets

Published on our datasets page

Oct 1, 2026

1.1B+
corpus tokens

SomNLP corpus from 17 sources

Sep 24, 2026

5
bootcamp cohorts

Delivered since September 2025

Oct 1, 2026

200+
bootcamp alumni

People who took part in a cohort

Oct 1, 2026

What we do

  • Data

    Open Somali text and speech corpora with documented sources and licences.

  • Models

    A Somali tokenizer, speech-to-text checkpoints and translation previews.

  • Education

    Free bootcamps and mentored learning tracks for Somali engineers and analysts.

  • Technology consulting

    Software and advisory work for partner organisations, including websites and mobile apps.

History

  1. 2023

    Somast founded

    Started in Mogadishu with a mission to build open AI infrastructure for the Somali language.

  2. 2025

    First open corpora released

    Published the initial Somali text and speech datasets under permissive licenses.

  3. 2025

    Talent programs launched

    Began fellowships and bootcamps training local engineers in modern ML and NLP.

  4. 2026

    Models & benchmarks

    Released ASR checkpoints, a base tokenizer, and Soplang, a compiler for the first Somali-syntax programming language.

  5. Next

    TTS, RAG & partnerships

    Somali speech synthesis, retrieval tooling, and regional research collaborations.

Programmes

  • Git & GitHub BootcampJuly 2026

    Version control from first principles, commits, branching, merge conflicts, pull requests, and the collaboration workflow every other bootcamp assumes you already know.

  • DS & ML Bootcamp - Jun 2026June 2026

    Hands-on Data Science and Machine Learning cohort with real datasets, model building, and capstone projects. Our largest cohort to date.

  • DS & ML Bootcamp - Feb 2026February 2026

    Second Data Science and Machine Learning cohort. Covered statistics, sklearn, deep learning fundamentals, and real-world Somali datasets.

  • Python for EveryoneApril 2026

    Beginner-to-intermediate Python bootcamp open to all. Our most-starred repository, stars and forks from learners around the world.

  • DS & ML Bootcamp - Sep 2025September 2025

    Our first full Data Science and Machine Learning cohort. Laid the foundation for all future programs with a curriculum now used by dozens of learners.

All bootcamps

Leadership and team

  • Ismail Ainte

    CEO

    Leading Somast's strategy, operations, and growth.

  • Sharafdin Yusuf

    Lead Engineer & Researcher

    Leading the design and development of Somast' technology platforms and research.

  • Omar Tood

    Lead Instructor & Researcher

    Data pipelines, machine learning models, and technical training programs.

  • Salah Sharafdin

    Software Developer

    Building web applications, user-facing features, and internal tools.

Meet the team

Advisors and board

Advisors and board members will be listed here.

    Partners

    • UnkadResearch partner of Somast.
    • DugsiiyeSponsored every Somast bootcamp, all open to anyone.
    • MemvidSomast was Memvid's lead open-source software engineer.
    • SoplangA separate open-source project under Somast: a programming language with Somali syntax.
    • DockerSponsors Soplang, our open-source Somali-syntax programming language, through the Docker-Sponsored Open Source program.
    • SkilfulSomast is Skilful's technology partner, providing consulting and software development.
    • Casri StoreSomast is Casri Store's technology partner and consultant, and built its website and mobile app.
    • Youth ConnectSomast is Youth Connect's technology partner and consultant, and developed its website.
    See all partners Become a partner

    Funding

    Somast is funded by three kinds of support.

    • Sponsorship: Dugsiiye has sponsored every Somast bootcamp, all of them open to anyone.
    • Client work: technology consulting and software development for partner organisations.
    • Open-source sponsorship: Docker supports Soplang through the Docker-Sponsored Open Source program.

    Press and recognition

    Press, talks, awards and publications will be listed here.

      Policies and openness

      We publish our terms and privacy practices, and release datasets, models and code under open licences.

      Contact

      Press and general
      [email protected]