eProsima is proud to announce the release of SustainML v0.2.0, a major update to its backend and developer framework for sustainable machine learning. This new version strengthens stability, usability, and extensibility across both the backend and the Qt-based front-end, introducing smarter task management, improved data handling, and deeper integration with research partners and frameworks.
The SustainML Framework is designed to provide an energy-optimized hardware solution and a corresponding machine learning (ML) model for a given user-defined problem, while taking into account the carbon footprint associated with training.
To support modularity, scalability, and interoperability, the framework relies on eProsima Fast DDS as its communication middleware. Fast DDS enables distributed components, referred to as Nodes, to exchange data efficiently and consistently across the system.
Researchers Eya Ben Chaaben and Janin Koch from INRIA and Université Paris-Saclay presented a forward-looking paper at the SIGCHI 2023 Conference on Human Factors in Computing Systems, held in Hamburg, Germany. Their study, “Addressing Sustainable ML Life-cycles through Human-Centered Design”, advocates for a human-centered rethinking of machine learning (ML) development practices, emphasizing sustainability not only during training but across the entire AI lifecycle.
At CHI 2025, researchers Nanna Inie, Jeanette Falk, and Raghavendra Selvan presented a timely and important study titled How CO2STLY Is CHI? The Carbon Footprint of Generative AI in HCI Research and What We Should Do About It. The paper evaluates the environmental impact of generative AI in human-computer interaction (HCI) research, based on an analysis of all 282 accepted papers from CHI 2024 that explicitly reported using generative AI.
The SustainML Developer Framework v0.1.0 by eProsima offers an intuitive way to define machine learning problems, customize goals and hardware, and run iterative experiments. It seamlessly collects and visualizes key sustainability metrics such as latency, power use, and carbon intensity, helping you compare results across models and devices. With easy setup via source or Docker and built-in integrations like Hugging Face model search, it provides a streamlined environment for sustainable AI experimentation.
SustainML is among these nine innovative projects dedicated to creating a sustainable ML framework for Green AI.
Plaza de la Encina 10-11, Núcleo 4, 2ª Pl.
28760 Tres cantos - Madrid (España)
This project has received funding from the European Union’s Horizon Europe research and innovation programme under grant agreement No 101070408.
