Centenum Labs is the research arm of Centenum Technology, exploring the frontier of neurosymbolic AI, symbolic regression, and program synthesis. We build systems that don't just predict. They understand.
Learning mathematical expressions directly from data. We explore SR as a bridge between raw ML and interpretable, human-readable models, especially for scientific domains.
Machines that write programs. We study how systems can infer compact, reusable abstractions from examples, moving toward machines that generalize like humans do.
Unifying the statistical power of deep learning with the precision of symbolic reasoning. We believe the next generation of AI must be able to explain itself.
Designing domain-specific languages that allow AI systems to build hierarchical representations. Structure as a first-class citizen in machine intelligence.
The Abstraction and Reasoning Corpus is one of AI's hardest open problems. We explore approaches that combine perception, analogy-making, and symbolic search.
Building tools and research that address real problems, from government systems and healthcare to language models for underrepresented African languages.
The dominant paradigm in AI is pattern matching at scale: powerful, brittle, and opaque. Systems that work until they don't, and fail in ways no one can explain. Centenum Labs was founded on a different conviction.
We believe the most durable AI systems will be those that combine the learning capacity of neural networks with the precision and transparency of symbolic methods. Not one or the other. Both, in tension, producing something new.
We sit at the intersection of fundamental research and real-world application. Our parent company, Centenum Technology, has shipped production AI systems for Nigerian banks, government agencies, and enterprise companies. That ground truth shapes everything we build in the lab.
We are building at the frontier, and we are building from Africa.