Smart farming at Lumboakandhoo

Planned · Q4 2026 · Lumboakandhoo, Maldives

Food production on small islands is a tight, honest problem. Land is limited, freshwater is precious, feed and fertilizer are mostly imported, and the climate margin is narrow. Getting more food from the same island — without quietly degrading it — is a question that rewards careful observation more than clever shortcuts.

This project is open research on island-scale smart farming, centered at Lumboakandhoo in the Maldives, in partnership with Big Fish Maldives, who runs the farm as a working site.

Partner and site

Lumboakandhoo is a working farm island maintained by Big Fish Maldives. It gives the project something rare for this kind of research: a real operating environment, with real animals, real soil, real weather, and real economic constraints — not a demonstration plot. Methods that earn their keep here have a chance of earning their keep elsewhere.

What we are working on

Poultry monitoring — using AI on low-cost cameras and sensors to track flock health, behaviour, feed and water intake, and early signs of stress or disease, so the people running the farm can act before small problems become large ones.
Soil and water — monitoring soil moisture, salinity, and irrigation patterns alongside the limited freshwater resources of a small island, and modelling how each one moves through the season.
Microclimate — temperature, humidity, and rainfall observed locally rather than inferred from distant stations, so that recommendations are tied to what is actually happening on the island.
Predictive support — short-horizon forecasts for irrigation needs, feed requirements, and yield, in a form farm staff can read at a glance.
Honest accounting — what each AI system is good at, where it is wrong, and how often it should be checked against a person who knows the animals and the land.

How AI is used

The role of AI here is narrow and useful: watching things that are tedious or impossible to watch continuously by hand, and surfacing the few patterns that matter. The decisions stay with the people running the farm.

Models are small, run close to where the data is collected, and are designed to keep working when connectivity is intermittent — which on a small island, it often is.

How we work

The farm is the expert. Our role is to bring open methods and modest tooling that make existing knowledge sharper, and to publish what works and what does not, so that other small-island farms can adapt the parts that suit them.

Data observed at Lumboakandhoo stays with the people who observed it. We collaborate without asking partners to surrender data rights to participate.

Themes this project anchors

Updates

We will publish findings, methods, and open datasets as the work develops. The first publications are expected later in 2026.