Two weeks doing the work of an AI Engineer, inside a lab that ships AI systems Caribbean banks and government agencies run on. One real project, one engineer as your mentor, one deadline. Run by StarApple AI, the Caribbean's first AI company, and Maestro AI Labs, a regional frontier AI lab. The Caribbean AI Association and Innovate 10X have already paid for your place. Filling in the form is the only part left to you.
Applications close Friday 7 August 2026
This is not a classroom course that ends with a certificate. You are given work that the lab needs done, and you do it. The Caribbean AI Association and Innovate 10X cover the cost of your place, so no family pays anything to take part. Every student is assigned their own engineer, which is what limits how many people can come.
You work alongside engineers who build AI systems for Caribbean banks, insurers and government agencies. On the first day you are given a written brief and one engineer who is responsible for helping you. You write code. It stops working. You find out why and fix it. Then you explain your choices to your mentor, which is what every engineer in the room does with their own work.
A place on this programme would normally be charged for. Every place in this cohort is covered by a scholarship, so there is no tuition, no deposit and no fee at any point.
Most teenagers use AI tools. Very few have built one. Two weeks is enough time to cross that line.
A sixteen-year-old has almost nothing to put on a CV, so nobody can check it. Work is different. Work can be opened, run and checked by anyone. You leave here with work.
AI systems that read documents, answer customer questions and write code are already running inside Caribbean banks, insurers and government ministries. Every one of them needs an AI Engineer to build it and an AI Leader to decide what it is allowed to do. Those roles are being filled this year. The people filling them in 2032 are choosing their first project now.
Millions of teenagers can ask a chatbot to write an essay. Almost none have had a senior engineer read their code, explain what is wrong with it, and watch them fix it. The second skill is the one an AI Engineer is paid for, and almost nobody your age has practised it. That gap is exactly how far ahead two weeks puts you.
You finish with three things: a project you can open on a laptop and explain in two minutes, a written reference from the engineer who reviewed your work, and two working professionals who know what you can do. Universities and employers can verify all three. Someone who spent the same two weeks on holiday has none of them.
AI models built elsewhere are trained mostly on data from elsewhere. They misread Caribbean names, addresses, accents and weather patterns, and those errors cost local companies money. Correcting them takes AI Engineers who live here and spot the mistake on sight. Right now the region has nowhere near enough of them, which is the whole reason this scholarship exists.
Nobody becomes either one in two weeks. What happens in two weeks is that you find out which of them you want, and you get the first piece of evidence that you can do it.
The person who builds the system and is answerable for whether it works. They write the code that connects a model to real data, they find out why it gives a wrong answer, and they fix it before a customer sees it.
What you do here: write code that a senior engineer reviews line by line, and ship it once it passes.
The person who decides what gets built, what does not, and who carries the risk when a system affects real people's money or health. Every Caribbean bank, insurer and ministry now needs someone who understands the technology well enough to say no to it.
What you do here: stand in front of the StarApple AI and Maestro AI Labs teams and defend the choices you made.
You say which one you want on the application form. We assign you in week one, based on what the lab needs done and what you are best at. Every project ends with software that works.
Build the software around an AI model that makes it useful. That means loading documents, breaking them into sections, searching those sections, and returning an answer with a link to the exact page the answer came from.
Example project: a search tool covering a few thousand pages of insurance policy documents that never gives an answer without showing the page it came from.
Work out whether an AI model is accurate enough to be used. You build the set of test questions, run the model against them, count the errors, and write the report that says whether it is ready.
Example project: a test set of Caribbean names, addresses and place names, and a scored report showing exactly which ones the model gets wrong.
Turn a working AI model into a screen that someone with no technical background can use without being shown how. You design it, build it, then sit with a real user while they try it.
Example project: a single-screen tool that lets a shop owner ask questions about their own sales records in plain English.
Work with the Maestro AI Labs team on a question nobody has answered yet. You design the experiment, record the results, and write up what happened, including the attempts that failed.
Example project: measure how much accuracy an AI model loses when it reads Caribbean English compared to American English.
Monday and Tuesday you learn the software the lab uses, how the team tracks changes to code, how the existing code is organized, and what your project is. Students start at very different levels and those two days exist to close that gap. From Wednesday you are building. By Friday your code is in a real project and running.
A fifteen-minute team meeting each morning and your mentor available all day. You finish the build, and this is where you learn that the difficult part is rarely the AI model, it is the messy data and the unusual cases. Your mentor reviews your code and you write the instructions someone else would need to run it. On Friday you demonstrate it to the StarApple AI and Maestro AI Labs teams and answer their questions about the choices you made.
StarApple AI builds AI systems that banks, insurers and government organizations across the region use every day. Their engineers review your code. They apply the same standard to your work that they apply to each other's.
Maestro AI Labs builds and studies AI models for Caribbean use, including research questions that no other organization in the region is funded to work on. Interns assigned to the research project work directly with this team.
Some interns arrive having written a lot of Python. Others arrive having written none. Experience at the start has not predicted who does well. Whether someone finishes what they start has.
No. Some interns arrive having written Python. Others arrive having never written a line of code. Week one is designed to bring everyone up to the same level. On the form, describe something you have built, organized or repaired. It does not have to involve computers.
Nothing. Every place is a full scholarship funded by the Caribbean AI Association and Innovate 10X, so tuition is covered in full. There is no application fee, no deposit and no charge at any point. If travel or equipment is a barrier for you, say so on the form and we will work it out.
Yes. The programme is open across the Caribbean. Most of the work is done remotely. The in-person days take place in Kingston, and if you live in Port of Spain, Bridgetown, Georgetown, Castries or anywhere else in the region, you join those days by video call instead. Tell us where you live on the form so we can plan for it.
A full working day, Monday to Friday, for two weeks. Some days finish earlier. Some run late when a deadline is close. If you already have a commitment in that period that you cannot move, write it on the form so we know before the programme starts.
If you are under 18, they confirm that you can take part and give us a contact number. There is nothing for them to pay. They can email info@beagenius.org with any question before you apply.
You keep the project and the written reference. Beyond that, past students have returned as mentors on later Genius Project programmes, been offered paid work by the labs, or used the project in their university applications. None of those outcomes are guaranteed, and all of them have happened.
It will not be read for this cohort. Applications close at midnight on Friday 7 August 2026 and offers go out four days later, so there is no window to add people afterwards. If you are still deciding, apply now and tell us afterwards if something changes. Withdrawing is easy. Getting added late is not possible.
A person reads every application. The two things that help you most are a specific answer to why you want a place, and clear evidence that you finish what you start. We do not filter on grades, and we do not consider what your family can afford. General answers that could have been written by anyone are what hold applications back.
Your application has been received. A person reads every one, and we will contact you by Wednesday 12 August 2026 whether or not you are offered a place.
If you are under 18, tell your parent or guardian to expect an email from us at the address you gave. Add info@beagenius.org to your contacts so it does not go to your spam folder.
Back to The Genius Project →Fields marked with a red star are required. The rest help us assign you to the right project.