Yali Labs Launches Beta of ALTA Foundry: Pioneering Local AI in Rwanda

AI Quick Summary
- Kigali-based startup Yali Labs has launched ALTA Foundry, a "Digital Refinery" designed to convert raw data into high-quality training data for AI, specifically addressing the challenges of building AI in Rwanda.
- Yali Labs is a Rwandan AI company focused on developing foundational AI models that understand and interact in the Kinyarwanda language, providing APIs for local businesses.
- ALTA Foundry tackles key data challenges by cleaning noisy audio and extracting machine-readable text from physical documents using OCR and transcription.
- Initially an internal tool, ALTA Foundry's beta launch opens its doors to the community, inviting contributions and feedback to ensure the AI technology accurately reflects local language and context.
- The platform aims to train multimodal AI models, starting with the banking sector using call-center audio, and expanding to include image recognition for local documents and signage.
Yali Labs continues its mission to build language-first AI products for Africa, focusing on Kinyarwanda foundational models and democratizing AI access.
This week, the Kigali-based startup Yali Labs announced the beta launch of its flagship platform, ALTA Foundry. While the global tech race often feels far away, this launch brings the "industrial machinery" of AI directly to Rwanda. It is a specialized "Digital Refinery" built to solve the single greatest challenge in the AI race: transforming messy, raw data into the high-quality training fuel needed for smart machines.
What Exactly Does Yali Labs Do?
To understand the Foundry, you have to understand the company. Yali Labs is a Rwandan AI powerhouse led by co-founders Philbert Murwanashyaka (CEO), Schadrack Niyibizi (Tech Lead), and Hirwa Gael (Operations Lead). Their team is not just building apps; they are architecting Sovereign AI.
Their daily work involves building foundational AI models that understand the nuances of the Kinyarwanda language. From healthcare to finance, Yali Labs provides the "brains" (APIs) that allow other businesses to create digital tools that speak, listen, and think like a Rwandan; not like a generic global model.
The Workbench: Why ALTA Foundry Matters Now
Since 2024, the team has been in the trenches, building Kinyarwanda data from scratch. They faced a wall where raw audio was often too noisy or slow to process, and legal hurdles made using external datasets a minefield.
ALTA Foundry was born to be the solution; a workbench where data is scrubbed, structured, and legally secured. It handles the heavy lifting:
- Audio Cleaning: Stripping background noise from local recordings so AI can "hear" clearly.
- Data Extraction: Converting physical books and PDFs into digital, machine-readable text using OCR and transcription.
Opening Doors to the Community
The biggest shift in this beta is the change in access. For two years, ALTA Foundry was a private tool used only by Yali Labs organizers. Now, the doors are open.
By moving away from internal-only use, the founders are inviting the community to step inside. People can now contribute to datasets, help refine AI models, and provide feedback; ensuring the technology truly reflects local language and context. It’s no longer just a startup's tool; it’s a platform built for the people to shape how their language is represented in the digital age.
Training a Multimodal Future
The real impact of ALTA Foundry begins where the cleaning ends. Yali Labs is now moving into the critical phase of training AI models, with the banking sector as the first major goldmine. Call-center audio provides thousands of hours of natural conversation, helping AI understand how Rwandans actually speak and interact.
But the roadmap goes beyond just sound; it’s about teaching AI to "see". The future of the Foundry lies in training models to recognize images; from handwritten documents to local signage; and instantly translate them into accurate Kinyarwanda text. By bridging the gap between sight and language, Yali Labs is ensuring that the digital world understands Rwanda in every format.
The engine is live, the mission is clear, and as Philbert Murwanashyaka puts it, the goal for everyone is now to "embrace and contribute." The era of truly Rwandan AI has officially arrived.
If you enjoyed this article, follow us on WhatsApp for daily tech updates. If you have an idea, need to be featured or need to partner, reach out to us at editorial@techinika.com or use our contact page.
Don't let the story end here.
Join 12+ others discussing this topic. Share your thoughts, ask questions, and connect with the community.

