I usually don’t write about things I haven’t done yet. It feels premature, like celebrating before crossing the finish line.
But this one is different. This article is about the preparation, and that part is already done. The hackathon itself? That’s coming in early February. Consider this part one of what will probably become a series.
How It Started#
About a month ago, it was just another Monday. Another weekly sync with my manager about “AI in engineering.” We’d been circling this hackathon idea for six months, discussing it in various forums, refining the concept, waiting for the right moment.
Then he said the words I’d been waiting for: “It’s time to put the hackathon into action.”
Game on.
From Idea to Approval#
First thing I did was crystallize the topic. After some deliberation, I landed on this:
Title: Build the Future: Autonomous Agents
The challenge: Design and deliver a multi-agent system that solves a real problem within a major feature. Build autonomous AI agents that can run without human intervention.
I drafted a single A4 page with high-level details, the kind of document that’s detailed enough to be taken seriously but concise enough that executives will actually read it. Took it to the higher-ups.
Got approved. Got a budget.
I was thrilled.
Getting the Word Out#
We have a company townhall every month. Perfect timing. I announced the hackathon there, got initial reactions, and started building momentum.
But an announcement isn’t enough. People forget. People get busy. You need a place where the hackathon lives, where people can engage with it continuously.
So I did what any reasonable person would do: I vibe coded a hackathon website.
The Hackathon Site (And Why It Worked)#
I built a dedicated site for the hackathon. Not just an information page, but an engagement platform. Here’s what it includes:
- Hackathon info: The basics, rules, timeline, what we’re building
- Teams: See what teams exist, who’s on them, join one
- Announcements: Updates and news as we get closer
- Resources: A curated collection of materials to prepare with
- Prep meetings: Sign up for sessions directly
- Betting: Because who doesn’t love some friendly competition? People can bet on who they think will win
- And more
But here’s where it got interesting: I added achievements and easter eggs. Everywhere.
Some were simple, like “Join a team” or “Sign up for a prep meeting.” Others were more complex and fun, like entering specific phrases into the site. One of my favorites: typing “I am not a robot” unlocks a hidden achievement.
This was a banger.
People started hunting for achievements and easter eggs. They explored every corner of the site. And on the way, they discovered the AI agent resources I’d curated. They saw which teams were forming. They registered for prep meetings. They engaged with the content.
They weren’t just informed about the hackathon. They were invested in it.
Side note: I’ve open-sourced a version of the hackathon site on my GitHub. If you’re planning something similar, I strongly recommend taking a look.
The Teaching Dilemma#
Here’s the challenge I wrestled with:
On one hand, I need to prepare the teams fast. The hackathon is in early February. I need to teach them how to create agents, how to improve them, how to make them work in multi-agent systems. Vibe coding is the key for speed.
On the other hand, I don’t want them to just tell Cursor or Copilot or Claude Code to create the solution. I want them to understand what they’re building. If they don’t understand the architecture, they can’t debug it. If they can’t debug it, they can’t improve it. If they can’t improve it, they lose.
But time is tight.
I tried to find a middle ground.
In the prep meetings, I focus on concepts and show the basics, maybe basics++, but not too much advanced material. Just enough for them to invest an hour or two a day playing with the tech and improving their skills on their own.
The goal: give them the mental models to learn faster, not do the learning for them.
The Prep Meetings#
From the start, we saw teams form quickly. Engineers got hyped. They started working on solutions before the hackathon even began, getting familiar with the technology.
As the days passed, the prep meetings helped engineers understand AI concepts much better, not just agents but the broader landscape: how models work, how to prompt effectively, how to evaluate outputs, how to architect systems that involve AI.
For one of the sessions, I asked our AWS Solutions Architect to join. He didn’t hesitate. He prepared an amazing two-hour deep dive on Strands and AWS Bedrock. The engineers loved it.
We still have about a week and a half before the hackathon. More prep meetings are scheduled. I’m also doing one-on-one sessions with engineers who want to go deeper, improve their skills, or consult about their hackathon approach.
The Real Goal#
Yes, the official objective is to build a multi-agent system. That’s what the engineers will be doing during the hackathon. But there’s a deeper reason we’re investing in this.
We could buy products like this off the shelf. Multi-agent frameworks exist. AI solutions are everywhere. But that’s not the point.
We want to train our engineers to have the skills that matter in the AI era. We want them to be in the top 1% of developers, not because they claim to be, but because we shaped them that way. Instead of hiring engineers who say they’re top 1%, we’re building a factory for creating them, exactly the way we need them.
This is what I do. I shape engineers for the AI era.
The hackathon isn’t just an event. It’s a forcing function for skill development. It’s a way to compress months of learning into weeks of intense, focused work. By the time February ends, our engineering team will understand AI agents at a level most developers won’t reach for years.
Why Stampli Invests in This#
This hackathon is a significant investment by Stampli. As Head of DevOps & Innovation, I get to wear two hats that complement each other perfectly.
The Innovation part is what allows me to bring my passion to the workplace, to push for things like this hackathon, to experiment with new approaches to engineering challenges.
The DevOps part is what allows me to execute it properly, to build the infrastructure, to create the systems that make it all work smoothly.
In future articles, I’ll talk more about other innovations I’ve brought to Stampli. Stay tuned for those.
The Final Touches#
As part of the preparations, I synced with our office manager and got her fully onboard. She’s working on an amazing theme for the day: swag, food, atmosphere. The kind of details that turn a hackathon from “a work event” into “an experience people remember.”
What’s Next#
The hackathon kicks off in early February. I’ll be documenting how it goes, what works, what doesn’t, and what we learn along the way.
If you’re thinking about running something similar at your company, here’s what I’d tell you: the preparation is where you win or lose. The hackathon day is just the culmination. By the time people show up, they should already be excited, already have teams, already understand the basics, and already be invested in winning.
That’s what the site, the achievements, the prep meetings, and all the engagement work is for. It’s not overhead. It’s the foundation.
Stay tuned for part two.
Building AI capabilities in your engineering org? Running your own hackathon? I’d love to hear how you’re approaching it. Find me on X or Telegram.