About pages are hard to write without sounding like an asshole. I'll try to be honest instead.
I work in cybersecurity. Have for a long time. I help enterprise security teams think through architecture -- identity, access control, Zero Trust, AI governance. I've had these conversations hundreds of times. I'm good at them.
Here's what I don't say in those conversations: I'm not always sure I'm right. The gap between what I recommend and what's actually true is something I think about more than I let on. The frameworks I cite were written by committees. The vendor capabilities I describe come from documentation, not from running the thing. There's a version of what I do that's well-credentialed guessing, and I've been honest enough with myself to admit that bothers me.
So I built it. Prometheus is a personal AI system running on a homelab K3s cluster in my house -- multi-persona orchestration, episodic memory, runtime authorization, autonomous build agents, affect sensing. The same architecture I'd spec for a client. Not a proof of concept. Production, running against my actual work.
Building it has been clarifying in ways I didn't expect. A personal AI system is a self-portrait. The gaps in my reasoning don't stay abstract -- they become broken services and wrong outputs. The things I care about become design decisions I didn't realize I was making. My blind spots show up in the system's behavior before I've noticed them in myself. You can't build something like this without it being deeply personal, and I didn't fully understand that until I was already in it.
This site is where I write about what I find. Original research on problems I think the field is getting wrong. Practitioner takes that I'd give to a peer over coffee but not in a vendor presentation. Build logs for how Prometheus works and what it cost me to get it there.
My thinking changes. I'll try to be honest when it does. I'm more interested in understanding things than being right about them, which sounds good on paper and is genuinely uncomfortable in practice. If something I've written is wrong, I want to know.
