Leveraging multi-textual representation of molecules on a large encoder-decoder foundation models for materials
- 2025
- ACS Fall 2025
Eduardo Soares is a Senior Research Scientist at IBM Research – Brazil, specializing in the design and pre-training of large multimodal foundation models for scientific discovery. His research focuses on integrating data-driven learning with physical modeling to develop AI systems capable of representing and predicting complex scientific phenomena across domains such as materials science, physics, and chemistry.
Previously, he contributed to the development of explainable AI techniques for autonomous systems and trustworthy deep learning methodologies. Eduardo earned his Ph.D. in Computer Science from Lancaster University, where his doctoral research received the Best Doctoral Dissertation Award from the International Neural Networks Society (INNS) in 2022. His work has been presented at leading conferences including NeurIPS, ICML, ICLR, and AAAI, and has accumulated over 3,000 citations.
Explore our work: IBM FM4M GitHub Repository