Fourteen years of enterprise IT, one printer fleet, and the same attack surface wearing a new protocol.
Short answer: I didn't leave IT to chase a trend. I'd spent 14+ years defending attack surface in banks, hospitals, and retail chains — then started building AI agents that touch real business tools. When I saw the same failure patterns show up in agents, I knew the market problem I've been solving for 14 years just moved layers. The 3D printing business taught me how to run one solo; agent security is where I've aimed it.
Enterprise IT leadership across regulated environments — financial services, healthcare, retail, higher education. The full record is on the about page; the short list:
That work all reduced to one discipline: map the attack surface, harden it, monitor it, re-test. It doesn't matter whether the protocol is WAN routing, store LTE failover, or a hospital EMR pipe. The job is the same.
Around the day job I built up a 3D-printing outfit — printers running Klipper and Bambu Lab hardware, a storefront, and the GaleTech YouTube channel where I talk about builds and the tooling around them. Running that shop is where I learned to ship fast without an enterprise safety net: you are the ops team, the support desk, and the SLA.
It's also where AI agents stopped being theory. Sourcing, customer email, content, research — one by one I found a workflow I could hand to an agent. Then I found the thing every builder finds: the moment your agent can call tools, you're not running a chatbot anymore. You're running software with hands.
Prompt injection, tool misuse, data leakage, over-broad permissions — I'd been solving these exact problems for 14 years, just at a different protocol layer. So I started red-teaming my own agents the way I'd red-team any production system: map the surface, attack it, prove the impact.
Then I took it to other people's systems — the same way I'd been doing it in enterprise. I've run public vulnerability work including a Khan Academy VDP writeup and published agent-security research, including CVE analysis from the live State of Agent Security benchmark. Bug bounty work keeps the skills sharp, and it keeps me honest: findings need reproductions, not adjectives.
Two things, stated plainly. First, you're buying a career of attack-surface thinking, not a prompt-engineering hobbyist's six months. Second, everything is flat-fee and published — the pricing page starts at a $149 snapshot, because I'd rather prove the methodology for $149 than ask anyone to bet $6,000 blind.
Tell me what your agent can touch. I'll tell you where it breaks.
Book the call →
Read next: What's Actually Inside a $6K AI Security Audit
Or try the free AI Prompt Scanner · ← Back to all posts