I run ifestos.ai. Architecture, perception, motion, and the long work of getting a machine out of the lab.
imperial college london·previously meta reality labs research·scape technologies·university of oxford
[01] work
Most engagements touch more than one, because robots rarely fail in only one place.
Sensor to compute to actuator. The decisions that are expensive to reverse, made early and made on purpose.
Vision, sensor fusion and learned models. Turning raw signal into something a machine can act on without guessing.
Planning, trajectory optimisation and control for machines moving through a world that does not hold still.
The distance between a prototype that works in the lab and a fleet that works in the field: integration, calibration, test, deploy.
[02] approach
A small core, with specialists from adjacent engineering disciplines brought in when a problem calls for them.
01
Before proposing anything we characterise what already exists — the hardware, the data, the failure modes. Most robotics problems are misdiagnosed before they are mis-solved.
02
A narrow system that genuinely works beats a broad one that demos. We aim for something measurable in the real environment, early.
03
Robotics problems rarely stay inside one discipline. When a project needs mechanical, electrical or controls engineering, we bring in collaborators who do that work properly rather than improvising it.
04
The engagement ends, the system does not. Code, documentation and the reasoning behind each decision are handed over, so your team owns it outright.
[03] who
I run ifestos.ai. Before that I built perception and robotics systems at Meta Reality Labs Research, Scape Technologies and the University of Oxford.
[04] contact
A description of the machine and the problem is enough to start. We will tell you honestly whether we are the right fit for it.