Case Study: Carbon Re reduces cement emissions with Weights & Biases

Carbon Re Develops Cutting Edge Decarbonization Technology with Weights & Biases

Carbon Re, a startup spun out from Cambridge University and UCL, faced the challenge of reducing the massive CO2 emissions from cement production, which accounts for 8% of global human-produced emissions. Their goal was to optimize the complex cement manufacturing process using AI but they needed a way to manage their machine learning experiments and models effectively. They turned to the vendor Weights & Biases for its ML platform to help institutionalize knowledge and accelerate research.

By implementing Weights & Biases, Carbon Re gained a centralized system for tracking experiments and model lineage, which streamlined team onboarding and knowledge sharing. The use of W&B Sweeps automated hyperparameter optimization, and W&B Models provided a reliable registry for managing deployments. This solution enhanced the team's productivity, allowing them to develop their Delta Zero Cement product, an AI agent that helps plant operators reduce fuel use and significantly cut carbon emissions.

Carbon Re

Theo Wolf

Machine Learning Engineer