Projects 6 min read

3rd Place at Hackathon Davos 2026 – Solo, Remote, AI-Powered

Last week I participated in Hackathon @Davos – the AI & Robotics Edition in the heart of the Swiss Alps. My result: An AI-powered dashboard for enterprise data migration, built in 24 hours, as a solo participant, from home. That was enough for 3rd place.

JIVS Migration Visual Companion Dashboard
The finished dashboard: AI recommendations, interactive charts, and a dark theme with enterprise-grade polish.

The Challenge

Hackathon @Davos is part of the Digital Lounge during the World Economic Forum. 24 hours to develop innovative solutions for real-world problems – typically in teams of 2-3 people.

My challenge came from JIVS, a provider of enterprise migration software. The problem: Their powerful migration engine produces complex, matrix-like data in Excel format. No problem for experienced consultants – but a real hurdle for new employees or stakeholders who need to quickly understand the status.

The task: Build a visual companion application that makes migration data understandable and interactive – without replacing the power of the existing solution.

My Approach: AI as a UX Accelerator

Instead of just building pretty charts, I wanted to go a step further. My vision: A dashboard that doesn't just show data, but explains it.

The core idea: AI-powered recommendations that automatically flag problems. The user doesn't have to dig through the data – the system tells them where things are broken.

Key Features

🤖

AI Recommendations

GPT-4o analyzes migration data and provides proactive hints like "High failure rate detected" or "747 items waiting for processing."

📊

Natural Language Visualization

Users can create charts via text input: "Show me failed records by object type as a pie chart."

📈

Trend Analysis & Predictions

Visualize historical data quality and generate AI-based forecasts for future developments.

📤

Multi-Format Export

Export data and dashboards as PDF, CSV, or JSON – for reports and stakeholder communication.

The Tech Stack

24 hours isn't much time. I relied on proven tools that I know well:

Component Technology
Frontend React 18 + Vite
Styling Tailwind CSS + Framer Motion
Charts Recharts
Backend Node.js + Express
Database SQLite
AI Integration OpenAI GPT-4o
Auth JWT

The result: A complete full-stack application with ~8,000 data records, user management, persistent dashboards, and AI features.

What Convinced the Jury

The evaluation particularly highlighted the code quality. At hackathons, you often see quick-and-dirty solutions – understandable given the time pressure. I deliberately focused on clean architecture:

  • Clear separation: Frontend, backend, services, routes – everything in its place
  • Documentation: README with setup instructions, API endpoints, architecture diagram
  • Production-ready: Docker Compose, environment variables, CI/CD pipeline
  • UX details: Tooltips, loading states, error handling, dark/light mode

Jury Feedback

"Impressive code maturity for a 24-hour project. The AI integration doesn't feel like a gimmick but brings real value for less experienced users."

Learnings: Solo Hackathon as a Remote Participant

It was my first time as a solo participant at a hackathon designed for teams. A few insights:

✅ What worked well

  • No coordination needed – I could set my own pace
  • No context-switching between different work styles
  • Full control over architecture decisions
  • Remote = no travel time, own setup, own coffee

❌ What was missing

  • No sparring partner for ideas
  • Fatigue hits harder without team energy
  • Less scope possible than with 2-3 people
  • The Davos atmosphere and networking

Conclusion

The hackathon reminded me once again: Under time pressure, the most interesting solutions often emerge. Sometimes you need exactly these constraints to focus on what matters.

If you have similar challenges – making complex data understandable, building dashboards for enterprise applications, integrating AI meaningfully – feel free to reach out.

Need to Make Complex Data Understandable?

I build dashboards and visualizations that non-technical users can actually understand. With or without AI.

Discuss Your Project