Hosted at The Circle · Enabled by Microsoft
Registration is managed through Climate Week Zurich.
A hackathon for AI engineers building intelligent solutions at the intersection of artificial intelligence and climate action. The EarthXCG AI Climate Hackathon invites anyone and everyone, no matter what their skills or background, to join and experience a day as an AI engineer during Climate Week Zurich.
Teams of 2-4 people have approximately 7 hours to tackle real-world climate challenges using the latest in large language models, machine learning, and intelligent automation. Build, test, and present a working solution that matters.
Work alongside like-minded engineers and climate innovators, learn from experts at EarthXCG and Microsoft, and compete for prizes while building something that could make a real difference in the fight against climate change.
Tuesday, May 5, 2026 · 08:30-18:00 CET
Check in, set up your workspace, and connect to Wi-Fi
Welcome from EarthXCG and Microsoft, overview of the day
Introduction to climate-AI challenges, tools, rules, and Q&A
Start building your solution
Take a 30-minute break, lunch provided on-site
All pull requests must be open by this time
5 minutes per team + 2 minutes Q&A
Winners announced across four categories
Thanks for participating!
Choose from three real-world climate problems to solve
Use Generative AI to make climate action accessible. Build tools for carbon coaching, climate Q&A, emissions advisory, or policy navigation.
Generative AIUse Computer Vision to reveal environmental insights. Detect deforestation, track solar adoption, map urban heat, or monitor climate infrastructure.
Computer VisionUse Predictive AI to optimize for carbon reduction. Forecast renewables, optimize building systems, schedule EV charging, or plan decarbonization.
Predictive AIClick any challenge above to read the full brief, including datasets, expected outputs, and judging criteria.
Get ready to build climate solutions on May 5
Read through the three climate challenges and start thinking about which one excites you most. Consider your team's strengths and interests.
Ensure your laptop is ready: install Python 3.8+, Git, and your preferred IDE. Test your API keys if you plan to use your own LLM services.
Teams of 2-4 work best. If you don't have a team yet, don't worry — there will be time for team formation at 09:20 on the day.
Bring your fully charged laptop, charger, and enthusiasm. API keys for LLM services are optional but recommended if you have your own accounts.
Fork or clone the GitHub repository to get started with the starter code and challenge briefs
Choose one of the three climate challenges that interests you and aligns with your team's skills
Use the starter code or bring your own stack. You have ~7 hours to build, test, and refine
Open a pull request before 16:45 CET with your code in submissions/<team-name>/
submissions/<team-name>/ folderSolutions evaluated across four weighted dimensions
Does the solution address a real, measurable climate problem? Is the potential impact meaningful at scale?
Is the solution working, well-structured, and technically sound? Does it make effective use of AI?
Is AI central to the solution — not just a wrapper? Does it demonstrate thoughtful model design or prompting?
Is the problem and solution clearly communicated? Can the team explain their choices and trade-offs?
Highest weighted score across all dimensions
Most meaningful climate solution
Most creative/effective AI application
Voted by participants
Anyone interested in AI and climate action! No matter your skills or background, you're welcome to join and build solutions during Climate Week Zurich.
Teams of 2-4 people are recommended. You can form teams before the event or join forces during the team formation period at 09:20.
Bring your laptop (fully charged), charger, and API keys if you plan to use your own. Enthusiasm and readiness to build something meaningful are essential!
Yes, lunch will be provided on-site around 13:00.
You can use any AI/ML technologies, frameworks, and APIs you prefer. Starter code is provided with common LLM integrations (OpenAI, Anthropic, Azure OpenAI), but you're free to use your own stack.
A judging panel evaluates solutions across four dimensions: Climate Impact (35%), Technical Execution (30%), Use of AI (20%), and Presentation (15%). Winners are announced in multiple categories.
Challenge details will be shared before the event for review, but all coding must happen during the hackathon (09:30-16:45 CET).
The EarthXCG organizing team will be available throughout the day for technical support, clarifications, and guidance.