An Applied AI Engineering Studio
Clients bring us a domain and a hard problem. We bring the stack — data pipelines, retrieval-augmented generation, multilingual models, and answers you can trace to a source — and ship a production system built around it. From Yuelu, China to Kigali, Rwanda.
Start a ProjectYUELU TECH LIMITED is an applied AI engineering studio, registered in Rwanda. Clients commission us to design and build production-grade AI products for their field — we own the engineering, they own the product and the brand it ships under. We also build and run projects of our own.
We take our name from Yuelu (岳麓) in Changsha, China — home to one of the world's oldest academies, standing for over a thousand years, and today to the Yuelu Summit at the frontier of AI and the digital economy. Scholarship and technology have always met here. We carry that DNA with us.
The same capability stack — ingestion pipelines, retrieval-augmented generation, multilingual understanding, and source-traceable output — adapts to whatever a client needs built, in whatever domain. The vertical changes; the engineering discipline doesn't.
Products we build for clients, and projects we run ourselves.
A multilingual rice-farming assistant for Rwanda — in English, Chinese, and Kinyarwanda. Retrieval-augmented generation grounds every answer in verified agronomic knowledge, from variety selection to pest and disease management.
ask.loudiagri.comA Chinese-language Douyin channel for Chinese companies entering Rwanda and East Africa — practical guides to company setup, payments, driving and housing, plus who's who in government and the latest appointments. Behind the leadership series: our own data pipeline, with a structured roster of officeholders checked against official sources.
Douyin @投资卢旺达One capability stack, re-applied to each client's domain. We build the parts that are hard to get right, and own them through to production.
The plumbing that feeds the model — collecting from public sources, APIs, and institutional archives, then cleaning, deduplicating, and structuring it into a retrieval layer the system can rely on.
A RAG architecture that grounds every response in verified, source-linked knowledge — keeping answers accurate, current, and traceable instead of hallucinated.
Natural-language understanding across English, Chinese, Kinyarwanda, French, and more — delivered on the channels users already live in, from web apps to messaging platforms like WhatsApp.
We start inside the client's domain — sitting with the people who will use the product to map real workflows, data sources, and decisions before a line of code is written.
We ingest and verify the domain's knowledge — public datasets, partner archives, an institutional source such as a national research body, a newsroom, or a regulator — into a retrieval layer grounded in fact.
We build the product and deliver it on the channel that fits — web, messaging, or API — then iterate against real usage data rather than assumptions.
We keep it running in production and extend it as the client grows. The same stack carries to new domains, languages, and markets — only the knowledge changes.
Tell us about your domain and what it needs to do. We'll tell you how we'd build it.
[email protected]Yuelu, Changsha · Kigali, Rwanda