Contents
Konstantin (Kostya) Ostrovsky is the Chief Architect at Torq, where he leverages over 18 years of experience in software engineering and architecture. He specializes in cybersecurity, with a background that began with writing Windows Kernel Drivers. Konstantin is also a frequent speaker at software engineering conferences globally.
At Torq, our goal is to be at the cutting edge of technology — both in how we build our products and in how we work day to day. We adopted GitHub Copilot early, rolled out org-wide access to ChatGPT, Claude, and Cursor, and coached PMs and engineers on promptcraft, coding with copilots, and fast iteration.
However, simply providing AI tools or just talking about them isn’t enough. While they might seem intuitive at first, mastering the art of working with AI takes time and practice. It’s a process of learning the tricks and developing an intuition to get the most out of them.
To accelerate adoption, we ran a single-day AI Hackathon designed to turn curiosity into muscle memory — what we call “vibe-coding”: to rapidly and intuitively build cool, product-related features using AI.
How We Ran the AI Hackathon
We gave teams permission to “vibe-code” — move from idea to working prototype in hours — without the friction of day-to-day priorities. The goals were simple: use as much AI as possible, build something useful or delightful, and learn repeatable patterns you can bring back to your sprint.
Sourcing and Filtering Ideas
We opened the floor to everything — product features, internal back-office tools, developer experience (DX) improvements, or anything else our teams could dream up. In a week, we collected nearly 40 ideas. We sat down with our colleagues from the Product Management team, who helped us filter the list by half, prioritizing ideas that were both fun to work on and valuable to our product roadmap. The R&D team selected the remaining ideas focused on internal tooling, DX, and other engineering priorities.
Forming Teams
With the list narrowed to 20 projects, we asked our engineers to vote for the ones they’d most like to work on. They could choose any project that interested them, even outside their usual domain. We voted on a few favorites and assembled balanced squads of 3-4 people, intentionally mixing collaborators who don’t often pair.
To help with this, I even vibe-coded a small Hackathon organization app. It optimized team assignments to ensure most engineers were placed on a project they had either suggested or voted for.
Creating the Atmosphere
HR and Finance went all-in: banners, shirts, an endless supply of food and drink, and an afterparty to keep the energy high. In true Hackathon fashion, it was also a competition. A jury of four well-respected representatives from different Torq departments awarded prizes to the top three teams, and a “Crowd Favorite” was crowned from a company-wide vote.
The Hack Day
Energy was high across offices, including Warsaw, where one new engineer who joined Torq the day before was able to contribute significantly and even took second place.
Everyone worked extremely hard and had a ton of fun. I was tracking the token usage on our AI tools, and the activity screen in Cursor and the buzz in the office showed teams working as late as 3am. Interestingly, some of the most active AI tool users were our Product Managers and Team Leads, not just the engineers.
The Big Demo
The next morning, teams got five minutes each to give either a presentation, PoC, or live demo in our preview environment. Some were fully functional projects that were live in our preview environment. The teams’ achievements were mind-blowing. The sheer volume of work, business value, and innovative concepts presented was astonishing. Projects that would normally take weeks were demoed after just 24 hours of focused “vibe-coding.” These weren’t production-grade solutions, but they gave everyone a powerful glimpse of what’s possible when leveraging AI tools effectively.
After the winners were announced, I sent a survey to all participants. The results were unanimous: Everyone had a fantastic time and found the experience incredibly valuable.
How It Went
Start planning early. Looping in the Product team upfront gave us well-thought-out problem statements and tasks, so teams hit the ground running.
We crowdsourced the roadmap. We asked everyone to submit ideas and vote. Ownership increased and teams landed on projects they actually cared about.
Encourage experiments. We explicitly allowed people to try new tools and approaches. The creativity and velocity that followed was off the charts.
Show the score. Mid- and end-event stats (such as progress, token usage, and demos shipped) got everyone pumped up, sparked friendly competition, and kept momentum high.
Next time, we’ll be sure to balance scope across teams. We’ll pre-size projects with a simple complexity rubric and right-size them at kickoff so every team tackles a comparably challenging task.
Want to Run an AI Hackathon at Your Company?
Here’s some tips and best practices I’ve learned from launching this initiative at Torq:
- Pick a single day.
- Open the funnel for ideas and filter for impact.
- Let people choose what they want to work on, then balance teams.
- Remove friction (e.g., tools, data, environments).
- Timebox. Demo. Celebrate.
- Ship the best two or three ideas into a productionization lane.
The cost was minimal for tokens, swag, and food. The ROI showed up immediately: reusable code, better AI workflows, and teams that left with confidence, not just curiosity.
So, what’s the key to driving AI adoption? For us, it was turning conversation into action. Torq’s AI Hackathon provided tangible proof of what our teams could accomplish, transforming abstract potential into mind-blowing demos. It’s the ultimate accelerator, compressing weeks of learning and experimentation into a single, high-energy day.
The challenge is to carry that momentum forward, integrating these new vibe-coding workflows into our regular sprints. This is how a one-day event becomes the foundation for a long-term, AI-native culture.
Love the idea of vibe-coding, AI Hackathons, and building the future of security automation? We’re looking for engineers, PMs, and problem-solvers who want to push the boundaries of AI-native development. Check out Torq’s Careers page and join us in shaping the future of security.
