Building Towards the Hackathon: A Month of Preparation
The hackathon starts this Friday. But for forty-five developers across fifteen teams, the real work began a month ago.
We knew from the start that a 48-hour event wouldn't be enough. You can't expect people to build sophisticated AI applications in 48 hours if they're also learning the fundamentals during that same window. So we made a decision: invest heavily in preparation, then let the hackathon be pure execution.
This is the story of what we built in the four weeks leading up to November 28th.
The Setup
Every Saturday for a month, participants gathered for intensive training sessions. Not lectures where people passively take notes, but working sessions where everyone built something real, received feedback, then returned the following week having tackled a challenge.
The structure was simple: teach on Saturday, practice during the week, ship something by Friday, then showcase what you built the following Saturday. Repeat for four weeks.
Between Saturday sessions, we hosted open talks every Wednesday with experienced AI practitioners. These weren't promotional webinars, they were candid conversations with people who've actually deployed AI systems in production.
Throughout the month, participants had access to a Discord community for discussions and questions. Qualified hackathon teams also had a dedicated WhatsApp group for coordination and support.
The preparation was structured yet flexible: Saturday sessions, weekly challenges, Wednesday conversations, and continuous community support. By the time teams arrive Friday for the hackathon, they'll have experienced a month of hands-on learning and building.
Week One: Getting Started
The first Saturday focused on foundations. Participants set up their development environments, learned how modern AI systems work at a high level, and got their hands dirty building something immediately.
We didn't want anyone leaving that first session without having created a working AI application. It didn't need to be sophisticated, it just needed to work. A text classifier. A simple generator. Something deployed they could point to and say, "I built this."
The week's challenge was clear: build your first AI application and be ready to present it by Friday.
When participants returned for the second week, some had completed the challenge, others were still working through it, but everyone showed up ready to learn. More importantly, everyone had questions. Real questions that only emerge when you're actually trying to build something and hitting obstacles.
That's when we knew people were engaged.
Week Two: Making It Yours
The second week centered on customization and fine-tuning. We explored how AI models could be adapted to different contexts, examining use cases in healthcare, education, and other domains.
The week's challenge encouraged participants to experiment with fine-tuning techniques and apply what they'd learned to their own ideas.
Some participants built and shipped projects. Others worked through concepts at their own pace. Our role was to deliver the sessions, share the knowledge, and create space for learning. What people did with it was up to them.
Week Three: Making It Real
The third week shifted focus to deployment. Building something on your laptop is one thing. Getting it running reliably so others can use it is another.
We covered what participants would need to deploy AI projects: running models locally and on cloud platforms, implementing systems capable of integrating external knowledge, adding voice interfaces.
Teams had already been formed before training began, so participants knew who they'd be working with for the hackathon. The week's challenge encouraged exploration of deployment concepts and integration techniques.
By the end of the third week, we'd covered the core technical concepts participants would need for the weekend ahead.
Week Four: Final Preparation
The final Saturday focused on applied AI and hackathon preparation. We discussed real-world use cases in education, healthcare, and finance, then opened the space for teams to work on their hackathon ideas with expert guidance available.
Some teams arrived with prototypes they'd been building. Others were still refining their concepts. Each team approached preparation differently, some built extensively, others focused on planning and research.
Throughout all four weeks, our role remained consistent: deliver the sessions, share the knowledge, and encourage teams to get their hands dirty preparing for the hackathon weekend. What each team did with that preparation was theirs to determine.
The Wednesday Conversations
Training wasn't limited to Saturdays. Every Wednesday, we hosted open talk sessions with AI practitioners who've built real systems. These sessions ran about 90 minutes, providing space for in-depth discussion.
Innocent Udeogu, CTO at Ubenwa, talked about building production AI systems that actually ship and scale, drawing from his experience in healthcare technology.
Ben-Manson Toussaint walked through how AI agents work and how to implement them practically, beyond theoretical concepts.
Dr. Rony Dupuy Charles led a session on building a prediction model from A to Z, what you need, how to build it, and how to deploy it. Practical, start-to-finish guidance.
Linh Pham, Co-Founder and CEO of Lexi, shared her journey from Harvard's Master in Design Engineering program to transforming clinical communication through AI-powered medical interpretation. Her insights on using AI to expand healthcare access were particularly resonant for our context.
These conversations weren't polished presentations. They were open discussions where participants could ask questions and learn from practitioners' concrete experiences.
All sessions were recorded so participants could review them later when concepts made more sense in context.
What Changed
By the end of four weeks, we'd delivered what we set out to deliver: four Saturdays of training, four Wednesday open talks, and continuous access to learning resources.
The questions evolved. Week one questions centered on basic setup and "how do I...?" By week four, participants were asking about optimization strategies and architectural choices. That shift demonstrated their engagement with the material.
Teams had been formed from the start, so participants knew their teammates throughout the preparation period. How each team used that time varied, some built extensively together, others focused on individual learning and planning.
The engagement from those who stayed through was genuine. People showed up, participated in sessions, and prepared in their own ways for the hackathon ahead.
Why This Mattered
Hackathons are often sink-or-swim events. You show up, figure it out, hope you can build something worthwhile before time runs out. Some teams succeed. Many struggle. Most of the weekend gets consumed by setup, learning, and debugging rather than actual building.
We wanted something different. A hackathon where teams could focus entirely on execution because they'd already validated their technical approaches. Where the 48 hours went toward building ambitious projects rather than figuring out basics.
The preparation created that possibility. When teams arrive Friday, they won't be starting from zero. They'll be starting from a foundation of knowledge, validated tools, tested deployments, and working prototypes.
The hackathon becomes an opportunity to combine and evolve what they've already proven they can build individually.
The Documentation Principle
Throughout this process, we documented what we could. Training recordings and open talk sessions are available for participants to review.
This wasn't proprietary training. We're sharing what we learned openly, hoping other communities can draw from our approach and adapt it to their own contexts.
We're building in public. Not because it's trendy, but because this work only has value when it's shared. The recordings are accessible. The community discussions continue.
Five years from now, someone preparing for a different event in a different place can look at what we attempted and learn from it. That's the goal.
The People Who Made It Possible
None of this happens alone.
This program exists because experienced practitioners chose to invest their time and knowledge into Haiti's developer community. Gueter Josmy Faure, Innocent Udeogu, Ben-Manson Toussaint, Jean Sauvenel Beaudry, Dr. Djinaud Prophète, Linh Pham, Patrick Selamy, and Dr. Rony Dupuy Charles, each brought years of expertise building real AI systems at Google, Meta, Harvard, and their own ventures. They didn't have to show up. They chose to.
Every Saturday session. Every Wednesday conversation. Every question answered on Discord. Every moment of guidance offered. That's not obligation, that's investment in something bigger than any single event.
The participants who showed up week after week deserve recognition too. Showing up consistently for a month isn't easy. Balancing learning with work and other responsibilities isn't simple. They invested their time.
We're building this together, organizers, mentors, participants, partners. Syntax Studio and Akademi provided infrastructure and support. Le Wagon Canada gave participants access to their Data/AI/ML learning platform and courses. But beyond the institutional backing, this worked because individuals believed it was worth their time.
This is how ecosystems start. Not with grand announcements or perfect conditions, but with people choosing to contribute what they can. We're taking this journey one step at a time, learning as we go, and inviting others to join as we build.
The door remains open. If you're watching from the sidelines and wondering if there's space for you in Haiti's AI community, there is. Whether you're just starting to learn or you've been building for years, whether you can contribute time or knowledge or just enthusiasm, we're building something that needs all of it.
We don't have all the answers. We're learning as we go. But we're learning together, and that's what really matters.
What Comes Next
This weekend, fifteen teams will spend 48 hours building.
They'll arrive having been through a month of preparation, four Saturday training sessions, four Wednesday open talks, and continuous access to learning resources and mentorship. Some built projects during preparation, others focused on absorbing concepts. Each team prepared in their own way.
The training provided the foundation. The open talks connected participants with experienced practitioners. The community created space for learning and support.
Now we discover what emerges when Haiti's developer community gathers to build together.
The hackathon starts Friday.
Let's see what they create.
Follow the journey:
