I build robots
that behave.

How much lab-grade capability survives a hard low-cost constraint? One coherent ROS 2 system on consumer hardware: 2D SLAM, a YOLOv8 perception pipeline, MoveIt 2 manipulation, and real-time motor control on an ESP32-S3 running micro-ROS / FreeRTOS. Validated per-subsystem — honest evidence over impressive video.

UPV had strong robotics students and no structure to aim them at international competition. Founded the association, grew it to 150+ members competing in the European Rover Challenge, and designed it to not depend on its founder — it keeps running while I'm 10,000 km away. That was the requirement.

The robots produced the data; the people who owned them couldn't read it. Built an API and visualisation named for what operators control, not how the system is built. Clients could act on their robots' state directly — the engineer left the loop.

A résumé as a hand-written canvas game — physics, particles, state machines, a cipher for the curious. The medium proving the message. The arm following your cursor right now is its quieter descendant.

• Thumbnail pipeline — branded SVG batch generation for a 70-day video series · Python
• Archive scanner — cataloguing tool for a 47,700-file project archive · Drive API
• Enclosure design — four-part printable chassis with interlocking joints · OnShape
• Repair bench — second-hand consoles, controllers and e-readers, revived