LinkedInยทTuesday, 4 August 2026ยท22d ago
๐ฅ Our AI for Earthquake Response solution is now open source. A few months ago, our team took 2nd place in the ESA ฮฆ-lab and Internationalโฆ
Eric Charles
CEO / Founder at Datalayer โก Managed AI Agents for Data Analysis โจ๐ค #faster #cheaper #collaborative
๐ฅ Our AI for Earthquake Response solution is now open source.
A few months ago, our team took 2nd place in the ESA ฮฆ-lab and International Charter AI for Earthquake Response Challenge, competing against more than 250 teams. Our solution combined deep learning, image registration, Siamese networks, and ensemble models to detect earthquake damage from satellite imagery.
Today, we're making the entire project publicly available.
๐ Read how we approached the challenge:
https://lnkd.in/eYieZisP
๐ป Source code:
https://lnkd.in/e-Bqbx5J
We believe competition solutions shouldn't end with a leaderboard. By sharing the code, we hope others can learn from it, reproduce the results, and build on top of our work.
This is only the beginning.
Over the coming weeks, we'll also publish a fully reproducible version on the Datalayer platform, allowing anyone to rerun the complete analysisโfrom data preparation to model training and evaluationโwith cloud GPUs and AI assistance.
If AI is going to help solve real-world problems, reproducibility should be as important as performance.
#AI #DeepLearning #SatelliteImagery #EarthObservation #OpenSource #Jupyter #DataScience #Earthquake #ComputerVision
โป 1
View on LinkedIn Cross-referenced
Related on the wire
Version 2 lays the foundation for ๐ฒ๐ป๐๐ฒ๐ฟ๐ฝ๐ฟ๐ถ๐๐ฒ-๐ด๐ฟ๐ฎ๐ฑ๐ฒ Jupyter MCP.
๐งญ ๐๐ฟ๐ผ๐บ ๐๐ถ๐ฏ๐ฒ ๐ฐ๐ผ๐ฑ๐ถ๐ป๐ด โ ๐ฑ๐ผ๐ฐ๐๐บ๐ฒ๐ป๐-๐ณ๐ถ๐ฟ๐๐ โ ๐๐ฝ๐ฒ๐ฐ-๐ฑ๐ฟ๐ถ๐๐ฒ๐ป ๐ฐ๐ผ๐ฑ๐ถ๐ป๐ด Iโve been using AI coding agentsโฆ
More to come... #next
This is a first step to give GPU access for code execution from your Agent. Thank you again Ryosuke TOKUAMI ๐๐
๐ ๐ฆ๐๐ป, ๐๐ฎ๐ป๐ฑ, ๐ฐ๐ผ๐ฑ๐ฒ, ๐ฐ๐ผ๐บ๐บ๐๐ป๐ถ๐๐โฆ ๐ฎ๐ป๐ฑ ๐๐ต๐ฒ๐ฎ๐๐ฟ๐ฒ. ๐ก๐ผ๐ ๐ฎ ๐ฏ๐ฎ๐ฑ ๐๐ฒ๐ฒ๐ธ๐ฒ๐ป๐ฑ. I came back from a few days atโฆ
Today is a meaningful milestone for Datalayer When we started building around Jupyter MCP Server, we had a simple belief: teams should notโฆ