HaitiAIGrant
On Base
GPU compute grant

Powerfor a risingApply nowgeneration

Haitian developers have the skill to train AI models and no way to pay for the GPUs. This grant buys the hours, and publishes every dollar it moves.

3 funded

See the projects

$500granted to 3 projects granted available

Still available

$251.80 on Base

aBasUSDC $204.60USDC $47.20

A100 compute

$1.20 / hour

0h
209h
Where the money went

$500 granted, project by project

See all projects
Other$150

Patrol-X: An AI-powered real-time event-monitoring system.

Ralph Djino DUMERA

Haiti

Agriculture$250

Crusaders - AgriBot

Falandy JEAN

Haiti

Education$100

Braina AI

Jens Joe Gladimyr Milfort

Croix-des-Missions, Port-au-Prince, Haïti

Granted to builders

$500

Projects funded

3

Still available

$251.80

Check the wallet on Basescan →
The gap

This is not a talent problem.
It is an access problem.

What they want to build

  • A Creole language model that understands local culture
  • AI tools for farmers, students, and small businesses
  • Health assistants for rural clinics
  • Smart systems to fix broken systems

What stops them

  • No access to AWS or Google Cloud
  • No local GPUs or infrastructure
  • No support programs for AI builders
  • No way to train — even when they know how

“They stay up late, not because they want to, but because it is the only time the electricity is stable enough to run a model.”

How a grant works

Four steps from an idea to a running job

  1. 01

    Tell us what you are training

    The application asks for the model, the data, and the hours you think it needs. Plain language is fine. A rough estimate is fine.

  2. 02

    We read it and ask questions

    Every application gets read by a person. If the compute plan looks thin, we come back to you before deciding rather than rejecting quietly.

  3. 03

    Credits land on a platform

    Approved projects get compute credits on whichever platform suits the job. Grants so far have run between $20 and $250.

  4. 04

    You ship and show the work

    Publish what you built — open source, a write-up, or a demo for the community. That is the only string attached.

Where the hours run

We do not own GPUs. We rent them by the hour.

A grant is not a cheque. It is time on someone else’s hardware, bought on the platform that fits the job.

Prototype

Google Colab

Notebooks with a free tier. Where most grantees start: shape the data pipeline, get the training loop right, fail cheaply before spending a grant on it.

Train

RunPod

On-demand GPU pods billed by the hour, including A100 class hardware. Where a run goes once the notebook works and it needs real memory and real time.

Anything else

Wherever the job fits

If a run needs hardware neither of those carries, we buy it where it exists. Tell us what the job needs in your application and we will price it.

What a dollar buys

100

compute units

$10

A100

GPU per hour

$1.20

8

GPU hours

per $10

Impact

This is just the beginning

Who we support

  • Students learning AI and ML
  • Hackers and creators
  • Developers building real solutions
  • People who want to build, not wait

What more hands unlock

  • Diaspora leaders
  • Tech mentors
  • Infrastructure sponsors
  • Cloud providers

Let’s make AI truly inclusive

We don’t need to wait for government, private sector, or big investors to step in. We can build AI with what we have.

Fund

Add compute credits or dollars to the pool.

Mentor

Give technical guidance to the teams building.

Share

Tell one person who should know about this.

Apply for a grant