The lab
A research lab
organised around
things that ship.
YTK Research works on intelligent machines and the systems around them. We are based in Emirates Tower, Dubai, and we operate four divisions under one engineering standard: YTK Robotics, YTK AI, YTK Spark and YTK Technology Consulting.
Not a consultancy
with a research page.
A lab with a business.
There is a difference between an organisation that studies a problem and one that has to make it work on a machine in the field. YTK is built for the second position. Our research agenda is set by deployment constraints: thermal limits, radio behaviour, inference latency, cost per run, and the tolerance of the person operating the system.
That is why the divisions are structured the way they are. Robotics and AI do the deep technical work. Consulting translates it into decisions an organisation can act on. Spark makes sure the result is understood by the people who need to buy it. None of the four is decorative — each one pays for its place in the structure.
We are deliberately specific about what we claim. If a capability is not something we have built and can demonstrate, it does not appear on this site.
Distinct identities.
Shared standards.
Each division has its own expertise, vocabulary and visual language. They share one research culture, one review process and one definition of done.
Five commitments
that shape the work.
These are not values on a wall. They are the rules we use to decide what to build, what to publish and what to turn down.
Demonstrable over impressive
If it cannot be shown running — on hardware, in a trace, in a field trial — it is not a capability yet. It is a hypothesis.
Constraints are the design
Thermal, power, bandwidth, latency and cost envelopes are not obstacles to the design. They are the specification.
Measure before you claim
Evaluation is part of building, not a stage after it. Numbers on this site are indicative lab figures and are labelled as such.
One lab, not four silos
The hardest problems sit between disciplines. A vision problem is often a radio problem wearing a different hat; we staff accordingly.
Hand over something operable
Every engagement should leave the client able to run what was built. Dependence on us is not a business model we want.
Start with
the problem.
Most conversations begin with a technical constraint nobody internal has time to chase. Describe it plainly — what has to happen, what is blocking it, what the deadline is. We will say whether it is ours to solve.
United Arab Emirates
Four divisions. One conversation to start.
Tell us what you are trying to build, and we will route it to the right people inside the lab.