Mission
To unify existing artificial intelligence and scientific computation systems on a multithreaded device.

Initiative
Current software solutions rely upon the task scheduler to perform parallel operations.  This is a derivative of the evolution of computation, but one that we need not suffer with.  Parallel architectures are currently available that would allow a simple way to virtualize serial execution asynchronously through coordinated timing.  The solution is simple - we oscillate a control bit at a logical frequency from which a standardized set of operations can be performed by a multitude of kernels on a GPU.  The implementation is only difficult as it requires a team of engineers capable of building a unified framework collaboratively.

Expectation
Our goal is to build a system from which many programs could leverage a shared logical space in GPU memory.  This would allow engineers to load shared data into memory for many operations to be performed at once, rather than each program being responsible for managing their entire worldview of the data.  In turn, this also allows us to build an intuitive space from which simulated environments could be created and calculated in real time to differing degrees - effectively allowing us to turn a single device into a simulated world tied to real world inputs.  As such we'd be able to construct sensors that are bounded to certain dimensions of physical space which could leverage machine learning to feedback into control surfaces.  This persistent feedback loop would effectively build the foundation for a unified foundation from which intelligent systems could be built by all of humanity.

Requirements
To do this we require at least two additional C programmers to iterate upon the idea, as well as a minimal amount of hardware.  At six months we should have a working prototype of our operating system and potential pathways for profit as a byproduct of creating of our tools.