Summary
Marco Siracusa is a Principal AI Performance Engineer with nine years of experience architecting AI supercomputers at the nexus of computer architecture and ML compilers. He has led design and integration of custom tensor marshaling hardware, developed performance models and compilers for embedding and recommendation workloads, and contributed to 13+ research projects with 11+ papers and two patents. Recognized as a 2025 Machine Learning and Systems Rising Star, Marco has held research and engineering roles at Arm, Samsung Semiconductor, Stanford, and the Barcelona Supercomputing Center, consistently moving ideas from prototype compilers to production-focused hardware. Based in Milan, he combines deep academic training (PhD-level work in computer architectures) with hands-on implementation of next-generation AI infrastructure—a rare blend that helps him optimize systems across software, compiler, and silicon boundaries.
8 years of coding experience
3 years of employment as a software developer
Master's degree Computer Science and Engineering, Master's degree Computer Science and Engineering at Politecnico di Milano
Bachelor's degree Computer Electronic and Telecommunications Engineering, Bachelor's degree Computer Electronic and Telecommunications Engineering at Università degli Studi di Parma
UPC Universitat Politècnica de Catalunya
Visiting Researcher Computer Science, Visiting Researcher Computer Science at Stanford University