An engineer renowned for making the open-source VLC media player handle video playback smoothly across countless devices is now turning that performance expertise to robotics -- the low-level software that governs how machines perceive, decide and move in real time. The throughline is a craft most people overlook: squeezing reliability and speed out of constrained hardware.
The move highlights a shift in where robotics' hardest problems actually live. As foundation models get better at high-level reasoning, the binding constraint increasingly becomes execution -- the unglamorous real-time control loops, latency budgets and edge cases that determine whether a robot is a smooth product or a jerky demo. That's a systems-engineering problem as much as an AI one.
“The move highlights a shift in where robotics' hardest problems actually live.”
For investors tracking physical AI, deep low-level talent migrating into robotics is a meaningful leading indicator. The category's winners will need both the brains and the boring, exacting performance layer underneath -- and the people who can build that layer are scarce. When a generational systems engineer picks robotics, it's worth noting where they think the value is going.