Donato Crisostomi

Donato Crisostomi is a 3rd-year Ph.D. student in Computer Science at Sapienza University of Rome, affiliated with the GLADIA research group under Prof. Emanuele Rodolà. He is also an ELLIS Ph.D. member and visiting researcher at the University of Cambridge.

• Ph.D. in Computer Science (current)
• M.Sc. in Computer Science (2021) from Sapienza University
• B.Sc. in Computer Science (2019) from Sapienza University

His research spans model merging, representational alignment, and universal representation spaces, with publications in top-tier venues like CVPR, NeurIPS, and ACL. His work includes cycle-consistent alignment methods, evolutionary model merging libraries, and training-free music generation techniques.

Recent article trends highlight expertise in Transformer optimization (2025), evolutionary model merging (2025), music generation (2025), and few-shot learning (2024). Key subfields include neural weight permutations, multimodal integration, and e-commerce NLP applications.

• Best Poster Award at EEML 2025
• Lead organizer of NeurIPS UniReps workshop

He has contributed to 15+ peer-reviewed publications and led the €1.5M NEXUS project's Model Reuse work package. Mentoring roles at LOGML 2025 and prior industry experience at Amazon Alexa/Search complement his academic profile.