The IMPACT AI Lab: Training the Next Wave of Caribbean AI Builders

Key takeaways

• IMPACT stands for Innovative Minds Pursuing AI and Creative Thinking, and it is run by maestro with the University of the West Indies and Section 9.
• The binding constraint on Caribbean AI is people who can ship, not access to models or compute.
• Builders train on shipping products such as Credit Garden and OYA AI, not toy datasets, under senior review.
• The PROMPTICA curriculum is the technical spine, with a heavy bias toward measuring whether a system actually works.
• Training in-region keeps both the user-side value and the builder-side value, the IP and equity, inside the regional economy.
• Students enter through PROMPTICA, professionals through intensives, and institutions through partnerships and sponsored cohorts.

Ask a Caribbean founder what is stopping them from shipping AI, and the answer is almost never the model. The model is free or close to it. The blocker is people. There is no shortage of brilliant graduates across the region; there is a shortage of people who can take a frontier model and turn it into something a Caribbean bank, ministry, or small business will actually deploy. That is a training problem, and it is the problem the IMPACT AI Lab was built to solve.

IMPACT stands for Innovative Minds Pursuing AI and Creative Thinking. It is the education and innovation hub maestro runs alongside the University of the West Indies and Section 9, our applied research arm. The premise is blunt. You cannot buy AI sovereignty off the shelf, and you cannot fly it in on a consulting contract that leaves when the invoice is paid. You have to grow the builders, in-region, on real work, and keep them here.

Talent is the constraint, not compute

The Caribbean produces strong technical graduates every year. UWI, UTech, and the region's other institutions turn out computer scientists, statisticians, and engineers who can hold their own anywhere. The trouble is what happens next. The best of them get recruited by firms in Toronto, London, and New York, often before they have shipped anything at home. The salary gap is real and no one should pretend otherwise. But the deeper loss is the one that compounds: every builder who leaves takes with them the local context that makes Caribbean AI work, and they take the IP they would have created here.

This is the brain-drain problem stated plainly. The fix is a credible path where a graduate can do interesting, well-supported work on products that matter, get paid fairly, and see their career compound in-region rather than abroad. Guilt and appeals to love of country will not build it. That path is what IMPACT is trying to build, one cohort at a time.

What the lab actually does

IMPACT is not a lecture hall with a new label on the door. It runs four things in parallel, and they feed each other. First, applied research: small teams take a live Caribbean problem, from Creole-language understanding to credit risk on thin-file customers, and push on it until there is a result worth publishing or shipping. Second, applied projects: builders work on maestro's real products, not toy datasets. Third, structured mentorship, where senior engineers and researchers review work week by week. Fourth, the PROMPTICA curriculum, which is the spine that holds it all together.

PROMPTICA is maestro's curriculum for building with large language models the way a working engineer does, not the way a prompt-tip thread suggests. It covers prompt and context design, retrieval, evaluation, agent orchestration, and the unglamorous discipline of measuring whether a system is actually getting better. It is opinionated about evaluation in particular, because the gap between a demo that impresses and a product that survives contact with real users is almost always a gap in how rigorously the team measures its own output.

Real products, not toy datasets

The fastest way to turn a student into a builder is to put them on something that ships. IMPACT does this directly. Builders rotate onto maestro products where their work reaches live users and carries genuine consequences.

Credit Garden is one of those products: an AI credit and financial-inclusion system aimed at the roughly six in ten Caribbean adults who are invisible to traditional bank scoring. Builders on Credit Garden work with alternative data, model fairness, and the regulatory reality of lending in multiple jurisdictions. OYA AI is another, a conversational and workflow product where builders learn to make an assistant genuinely useful inside a Caribbean business rather than impressive in a demo. You can see the full lineup on the maestro products page. Working on these systems teaches things no course can: how to handle messy local data, how to ship under a deadline, how to debug a model that behaves in production in ways it never did in testing.

The UWI and Section 9 partnership

IMPACT works because of who stands behind it. The University of the West Indies brings the academic depth, the research culture, and the reach across campuses and territories that no startup can replicate on its own. Section 9 brings the applied edge: it is where maestro's harder research questions get worked out and where the products draw their technical backbone. maestro sits in the middle, connecting the university's pipeline of talent to live commercial problems and the funding to support them.

That triangle matters. A pure university programme tends to stop at the paper. A pure company programme tends to skip the foundations and burn people out on delivery. Putting the two together, with a research lab between them, gives a builder both the grounding and the shipping experience in the same place. The graduate who comes through it has a publishable result and a product in production, which is a rare combination anywhere in the world.

From student to shipping builder

The pipeline is deliberately concrete. A participant usually enters through PROMPTICA or a UWI course tied to the lab, where they learn the fundamentals and build a small but complete system end to end. The ones who show aptitude and appetite move into a project rotation, paired with a mentor and placed on a real product team in a junior capacity. There they own a slice of something that ships, under review, with the safety net of senior eyes on their work.

From there the path forks in good ways. Some builders are hired directly onto maestro product teams. Some join partner companies across the region that are hungry for people who can actually deliver AI. Some take what they have learned and start their own ventures, now with a network and a portfolio behind them. The point is that every fork keeps the person, and their output, inside the Caribbean economy.

A week in the life of an IMPACT builder

The abstraction collapses fast once you watch a single week. Take a second-year UWI student, call her Shanice, who came through a PROMPTICA module and earned a rotation slot on Credit Garden. Her Monday starts with a stand-up where the product lead walks through what shipped over the weekend and what is broken. She picks up a ticket: the model is over-rejecting applicants from one parish whose income arrives in irregular cash deposits, the kind of pattern a salaried scoring model reads as risk when it is just how a higglering business runs.

Tuesday and Wednesday she is in the data, not in a slide deck. She pulls a sample of the rejected applications, sits with a senior engineer for an hour, and learns to build an evaluation set that isolates the failure rather than papering over it. Thursday she ships a change behind a flag and writes the test that proves it moved the metric in the right direction without breaking three others in the process. Friday is review: her mentor tears the work apart in the kind way that teaches, and what survives goes into production where a credit officer in May Pen will lean on it the following week. By the end of the month she has a commit history on a product that handles money, a measurable improvement with her name on it, and a habit of proving claims instead of asserting them. That habit is the entire point. It is also the thing no lecture transfers and no imported contractor leaves behind. For more on how that scoring engine is built, see our piece on physics-informed credit in the Caribbean.

What changes the brain-drain maths

Brain drain is usually framed as a salary story, and salary is part of it, but the number that moves a career is rarely the starting one. A graduate weighs the whole arc: what will I be working on, who will I learn from, will my output compound into something I own, and is there a ladder above me. For decades the honest answer in the region was that the interesting work, the senior mentors, and the equity all sat abroad. So the talented left, and each departure made the next one more rational.

IMPACT attacks that arc rather than the single salary figure. A builder who ships on Credit Garden or OYA AI is doing work as interesting as anything in a foreign lab, with mentors who have actually shipped, and a portfolio that translates into a maestro role, a partner-company hire, or their own venture with a network already attached. The illustrative figure of roughly four in five graduates staying in-region comes from the maths changing, not from patriotism: when the best path forward exists at home, staying stops being a sacrifice. Multiply that across cohorts and the region stops exporting the very people who would build its AI. That logic sits at the centre of the broader case for Caribbean AI sovereignty.

What employers and governments get

The pipeline only holds if the demand side is real, and it is. A bank, insurer, or ministry that hires an IMPACT-trained builder is not getting someone who can recite a model card. They are getting a person who has already debugged a model in production, already handled messy Caribbean data with all its gaps and quirks, and already worked under the regulatory constraints that govern lending, health, and public records in the region. That shortens the distance between a hire and a working system from quarters to weeks.

For governments the calculus is sharper still. A public institution that builds with regional talent keeps its data, its decisions, and its institutional knowledge inside its own jurisdiction, instead of routing all three through a foreign vendor who owns the resulting system. Sovereignty over public AI is not an abstraction when the model in question is deciding who gets a benefit or how a record is read. Employers and partners can open a direct line through the contact page or get started to sponsor or hire from a cohort, and the wider argument for keeping this capacity onshore runs through our look at the people building Caribbean AI.

Why training in-region keeps value at home

Every AI product carries two kinds of value. There is the value to the user, and there is the value that accrues to whoever built it: the IP, the data advantage, the equity, the institutional knowledge. When a Caribbean problem is solved by a team flown in from abroad, the first kind of value might stay, but the second kind leaves on the same plane the consultants do. Train the builders in-region, on local products, and both kinds of value compound at home.

This is the whole argument for AI sovereignty reduced to its core. Sovereignty is people who understand both the models and the place, not only a server in a data centre, though the server matters too. A model fine-tuned on Caribbean data by an engineer who has never set foot here will miss things an engineer who grew up in Spanish Town would catch in a sentence. IMPACT is the mechanism for producing that second kind of engineer at scale, and for keeping the IP they create inside the region's balance sheet rather than someone else's.

How to join or partner

If you are a student or early-career builder, the door is the curriculum. Start with PROMPTICA, ship a complete system, and put yourself in front of the mentors who staff the project rotations. If you are a working professional, the lab runs intensives designed to move you from competent coder to someone who can ship AI under real constraints. If you are a company, a university, or a public institution, IMPACT partners on applied projects, sponsors cohorts, and places builders. The fastest way in for any of these is to start a conversation through the lab directly.

The Caribbean does not have to choose between exporting its talent and building its own AI. With the right pipeline it can do the second instead of the first. Explore the IMPACT AI Lab to see the current programmes, partners, and products, and get started to apply, partner, or sponsor a cohort. The next wave of Caribbean AI builders is already in training. The only question is whether you are in it or watching it.

Frequently Asked Questions

What is the IMPACT AI Lab?

The IMPACT AI Lab (Innovative Minds Pursuing AI and Creative Thinking) is the education and innovation hub run by maestro together with the University of the West Indies and Section 9. It trains Caribbean builders in-region through applied research, mentorship, the PROMPTICA curriculum, and rotations onto real products, with the goal of keeping AI talent and intellectual property within the region.

Why does the Caribbean's AI future depend on talent before technology?

Frontier models are widely available and inexpensive to access. What is scarce is people who can turn those models into products that work for Caribbean institutions and understand the local context they operate in. You cannot import that understanding on a consulting contract, so the region has to train its own builders. Talent, not compute, is the binding constraint.

What is PROMPTICA?

PROMPTICA is maestro's curriculum for building with large language models the way a working engineer does. It covers prompt and context design, retrieval, evaluation, and agent orchestration, with heavy emphasis on rigorously measuring whether a system is actually improving. It forms the technical spine of the IMPACT pipeline.

What products do IMPACT builders work on?

Builders rotate onto maestro's live products. Credit Garden is an AI credit and financial-inclusion system for the large share of Caribbean adults invisible to traditional bank scoring. OYA AI is a conversational and workflow product. Working on shipping products teaches builders to handle messy local data, deliver under deadline, and debug models in production.

How does training in-region help keep value in the Caribbean?

Every AI product creates value for the user and separate value for the builder: IP, data advantage, equity, and institutional knowledge. When outside teams build Caribbean solutions, that builder-side value leaves with them. Training builders in-region on local products keeps both kinds of value, and the resulting IP, inside the regional economy.

How do I join or partner with the IMPACT AI Lab?

Students and early-career builders enter through the PROMPTICA curriculum and project rotations. Working professionals can take the lab's intensives. Companies, universities, and public institutions can partner on applied projects, sponsor cohorts, or place builders. Visit the IMPACT AI Lab page and the get-started page to apply or open a conversation.

How does the UWI and Section 9 partnership work in practice?

The University of the West Indies supplies the academic depth and the pipeline of graduates across its campuses. Section 9, maestro's applied research arm, works out the harder technical questions and gives the products their backbone. maestro sits between them, connecting the talent to live commercial problems and the funding to support the work. A builder who comes through it leaves with both a publishable result and a product in production.

Does the IMPACT AI Lab help reverse Caribbean brain drain?

That is its central aim. Brain drain is rarely solved by appeals to stay home; it is solved by making the best path forward exist in-region. By putting builders on interesting work with senior mentors and a portfolio that converts into a job or a venture at home, IMPACT changes the career maths so staying is the rational choice. An illustrative four in five graduates remain in-region.

What does an employer get from hiring an IMPACT-trained builder?

A hire who has already shipped under review, debugged models in production, and handled messy local data within Caribbean regulatory constraints. That compresses the time from hire to working system from quarters to weeks. For governments, building with regional talent also keeps data, decisions, and institutional knowledge inside their own jurisdiction rather than a foreign vendor's.

How long does it take to go from student to shipping builder?

It varies by aptitude and prior experience, but the path is deliberately fast and concrete. A participant learns the fundamentals through PROMPTICA, builds a complete system end to end, then moves into a mentored rotation on a product team where they own a slice of something that ships. Many are contributing to production within a single project cycle.

Is the IMPACT AI Lab only for computer science graduates?

No. The lab takes strong technical graduates from computer science, statistics, and engineering, and it also runs intensives for working professionals moving into AI. What matters is aptitude, appetite for shipping, and a willingness to measure your own work honestly, more than a specific degree on paper.