Summary
Francisco Mendes is a Senior Performance Modelling Engineer with nine years of hands-on experience at the hardware-software intersection, currently solving frontier AI deployment challenges at OLIX after a substantive CPU design and verification career at Arm. He combines production CPU microarchitecture and memory-system verification expertise with low-level C/C++ and GPU programming, focusing on making performance predictable across latency, bandwidth, and coherence constraints. His work spans cycle-accurate modeling, RTL verification, and profiling of real ML training workloads on GPUs (CUDA/HIP), and he routinely builds small experiments to isolate and explain performance effects. Notably, his master’s research pushed GPUs beyond conventional DVFS limits to improve energy efficiency for deep learning, reflecting a practical taste for crossing abstraction boundaries. Based in Cambridge, he pairs rigorous engineering discipline with a knack for translating hardware realities into software-aware system designs.
9 years of coding experience
7 years of employment as a software developer
10th - 12nd Sciencies and technologies, 10th - 12nd Sciencies and technologies at Escola Secundária Pedro Alexandrino
8th Degree - Curso Suplementar Trombone de varas Composição Musical Orquestra História da Música; Produção e Formação Musical, 8th Degree - Curso Suplementar Trombone de varas Composição Musical Orquestra História da Música; Produção e Formação Musical at Conservatório de Música D. Dinis
Erasmus Embbeded Systems, Erasmus Embbeded Systems at Eindhoven University of Technology
Master of Engineering - MEng Eletrotecnical and Computer Engineering, Master of Engineering - MEng Eletrotecnical and Computer Engineering at Instituto Superior Técnico
Portuguese, English