Crespo-Valero, Pedro

A multilingual and multicultural engineer who uses science, technology and humanities as lenses for understanding the world. Electrically trained and numerically inclined, Pedro's career traces a continuous arc from microwave/RF circuit simulation to bio-electromagnetics, through building a cloud-native simulation platform at scale, and now into the AI era of engineering. The connecting thread throughout: keeping the application in view – how the software fits the physics, and how the code serves the science.

At UPM Madrid (MSc Electrical Engineering, 2001) he wrote his first simulators (Prof. Page - ETSIT), developing both circuital approximations and full-wave mode-matching models for high-frequency devices. That early habit of translating physical intuition into working code set the direction for everything that followed.

His PhD at EPFL (Prof. Mosig group, 2002–2007) focused on Method of Moments formulations for laterally-bounded microwave structures. The centrepiece was MAMBO, a dedicated EM solver built in collaboration with ESA-ESTEC (Dr. C. Ernst) to simulate realistic space-grade planar microwave circuits. It was his first simulator with genuine professional ambition: a tool designed not just to validate theory but to handle the kind of multilayer, boxed structures that actually fly. The MMCODEF network provided an international research stage alongside it, and a brief RF-MEMS collaboration closed the doctorate with an unexpected detour into cleanroom fabrication, a memorable experience for someone whose instincts had always been numerical.

A decade at SPEAG and collaboration with ZMT (2007–2017) turned academic foundations into professional software for the telecommunications and medical industries. Contributing across the full stack to SEMCAD X and Sim4Life taught him that serious software is never just about the physics: from numerical core to user-facing layers, he learned what it takes to ship tools that others depend on. The FDA-cleared IMAnalytics for MRI RF safety compliance added high-stakes algorithmic work to the portfolio. The most rewarding thread was cDASY6 5G, a robotic dosimetry system for 5G devices whose core challenge, reconstructing fields from sparse measurements, blended simulation and measured data in genuinely novel ways and brought him into direct collaboration with measurement engineers and the robots doing the work. Alongside all of it, he provided technical support to IEEE and ICNIRP working groups on RF safety methodology, contributing code and analysis behind the scenes.

In late 2017, Pedro joined IT'IS Foundation and the work shifted to the cloud. As a founding engineer on o²S²PARC (the open-source computational core of the NIH SPARC program, osparc-simcore), he made a deliberate transition into systems and platform engineering: distributed microservices, multi-tenant cloud architecture, REST/OpenAPI contracts, containers, async Python services, and the operational infrastructure to sustain a live research platform. SPARC maps the peripheral nervous system to advance bioelectronic medicine, a genuinely international effort spanning research groups across the USA, Switzerland, and New Zealand. o²S²PARC is its open simulation backbone, providing the modular "Lego bricks" that researchers assemble into reproducible workflows. By early 2025 the platform serves 1,800+ users and hosts 80+ reusable computational services. → Platform paper (2025)

Currently he is back at the simulation layer, this time with AI as the new variable. On the one hand, actively experimenting with AI-driven development – agentic coding workflows, LLM-assisted design, autonomous development tooling. On the other, interested in how AI capabilities – surrogate models, intelligent parameter search, data-driven augmentation – can be embedded directly into simulation pipelines to accelerate research.