> making actual things ..

ahh, am trying to do that - largely self-funded RnD over 4 years -

Also trying to wake people up about the scope for ML to be applied to real engineering problems / processes ..

We hear of applied-RL/ML examples at the elite end, such as taming fusion plasma, weather prediction on a desktop, solving protein structure from DNA sequence etc..

Yet there are vast applications such as training robot hands to sort refuse for recycling, optimal heat flow in industrial processes, silicon chip layout, 2D photos to 3D models ...

My own startup domain is detecting geometry in pointclouds .. which is currently a lot of repetitive manual labor done for the construction industry.

It is frustrating talking to VCs .. because they really dont believe there is any other AI than LLMs.

My guess is the only people who will fund this kind of useful tech are post-exit technical founders.