About
Built where the exclusion was lived
Birtal Lab is a Sudanese research initiative building speech and language data for Sudan’s languages, with the communities who speak them.
Mission
To give Sudan’s languages a voice in the age of AI, by building speech and language data with the communities who speak them — on terms those communities set.
Vision
A future in which AI understands African languages as a matter of course, and the communities who speak them decide how the data that represents them is used.
Why Birtal?
A birtal is a Sudanese tray woven from palm fronds and straw, used to cover and hold food at gatherings. Family and neighbours sit around it to share a meal — it stands for home, generosity and sharing: it gathers and protects what is on the table.
That is what we do: we gather and protect a community’s voices, and turn them into knowledge that serves the people they came from.
From local knowledge to global research.
Our values
Consent first
Nothing is recorded without consent, and nothing leaves the foundation without a deliberate, recorded decision.
Reviewed by speakers
Every contribution is checked by other speakers of the same language before it counts.
Open methods
We publish our methods, measurements and limits, so results can be checked and built upon.
Built with, not extracted from
The people whose voices make the data decide the terms on which it is used.
Saying where we are
We are early, and we say so — including what we have not proven yet.
Founder & origin
“I lived the exclusion these languages now face.”
Birtal Lab was founded by a Sudanese engineer who began programming in 2017, in his second year of university — then spent over six years under a digital embargo that locked Sudan out of the global tech economy.
He entered competition after competition; the embargo shut most doors. In 2021–2022 he earned a place in the Google Africa Developer Scholarship (Android track, with Andela) and completed it despite the constraints — yet for years he could not publish a single app to Google Play, because accounts and payments were blocked.
In 2021 he joined his local IEEE community, which opened doors the embargo had closed. He volunteered actively for five years, organised many events, and today serves as an IEEE grant reviewer. Only in 2025 — after war forced his displacement from Sudan to Egypt — could he finally publish his first app and work as a developer without restrictions. He attended Google I/O 2024 (Egypt), built a knowledge network, and entered the global AI community.
The spark for Birtal Lab came from practice, not theory: during a hackathon he built a language-data-collection lab, and seeing the gap firsthand revealed the need to build it properly — as real research infrastructure. The same embargo that silenced a developer for six years is silencing entire languages forever. Birtal Lab exists so that it stops.
Background & journey
Birtal Lab grew out of years of building technology for African communities — and out of a simple belief: the people best placed to build for these communities are their own.
The founder, Eng. Saifuddin Ahmed, an engineer from Sudan, has spent years working where technology meets community — humanitarian and civic tech, tools for underserved regions like Darfur, and hands-on machine learning. When war displaced him, he kept building — training small models on limited hardware, offline, so they'd work even without the internet.
That practical work is what surfaced the real problem. Building a farming assistant for Sudan (AgriGuard, part of Google DeepMind's “Agents for Good”), it became clear the deepest barrier wasn't the models or the connectivity — it was that AI simply doesn't understand people's languages. The languages are alive; the data, the tools, and the research are missing.
So the mission grew wider than speech alone. It's about building the whole missing layer for these languages: open data and models, keyboards and fonts so they can be typed, dictionaries so they exist online, and — in time — tools for health, education and farming that finally work in people's own tongue. And building it so the community owns it.
Grounded in the African AI community through the Deep Learning Indaba and Masakhane, Birtal Lab is the founder's focus today: turning local knowledge into technology that serves — and belongs to — the communities it comes from.
Governance
Who owns the data
Contributing communities retain ownership of their language data; Birtal Lab is a steward, not an owner. See Data Rights & Consent.
Open model weights & methods
Where we train or release models, methods and benchmarks are published openly; openness does not transfer community ownership of the underlying data.
Safety commitment
Our Constitutional AI safety is grounded in the African Declaration on Internet Rights and Freedoms (AfDec). We will not knowingly enable surveillance or culturally harmful use.