It started with a napkin sketch at a coffee shop in Marina del Rey. Robert Felderman was a theoretical computer science grad student at UCLA when his advisor sent him to meet Danny Cohen, a legendary figure at USC’s Information Sciences Institute.

What followed was less an interview and more a torrent of ideas—drawn in ink, fired with excitement, and driven by a belief that networks could move 100 times faster than they did. At ISI, Felderman found a place where experimentation was currency and building things mattered more than publishing papers.

Alongside Cohen and collaborators from Caltech, he helped develop breakthrough networking technologies that would lead to the Atomic LAN and the founding of Miracom, a startup that quietly helped lay the groundwork for how modern data centers work today.

In this wide-ranging conversation, Felderman revisits the collaborative, open-source ethos of ISI in the 1990s, a time when researchers weren’t chasing unicorns, but ideas. It’s a story of fast networks, faster minds, and the kind of scientific culture that made progress possible.