I've met maybe a dozen people who really get what I'm trying to do.

That's enough.

They didn't show up through networking or luck. They showed up because I went deep enough into something specific that I became findable. Depth is a signal. The further down you go, the more precisely you describe yourself, and somewhere down there is a person who recognizes the description.

In AI, mechanistic interpretability is one of those places. Go far enough in and you stop running into people who are curious about AI and start running into people who are obsessed with the same narrow, strange questions you are. Same thing in programming languages, where the compiler nerds find each other in the margins of papers nobody else is reading. In systems work, the people who care about memory architecture and compute locality at the same time form their own quiet guild. In photography, the person who has spent years studying light in a single kind of environment knows exactly who else has done that. In music, the ones chasing weird tuning systems or forgotten genres that shaped the ones everyone knows, they find each other too.

The surface doesn't have any of this. It's crowded up there, and everyone is sampling widely, staying current, having opinions on things that are trending. It feels like keeping up. It mostly isn't.

So build the thing that solves your own problem. Make the thing you actually want to exist. Put it out there without sanding off the strange parts.

Your people are out there waiting to meet you.