Foundation Models & Document Intelligence
Study low-resource language, speech and vision, including how reliably models read documents and flag uncertain results.
We develop, test and translate intelligent systems through research, human evaluation and industrial collaboration.
Study low-resource language, speech and vision, including how reliably models read documents and flag uncertain results.
Study how people evaluate tool-using agents, set permissions and intervene when a task needs human judgement.
Study agricultural and environmental change through remote sensing.
Explore scientific machine learning, digital twins and optimisation.
Design studies, recruit participants and evaluate AI systems with people in Southern Africa.
Visit NareGrid (opens in a new tab)Explore our published conference abstracts, including research on identifying crop pests and diseases from imagery.
PublicationsFrom experiments to evaluated models to production-ready intelligence.