I'm 26, a Senior Manager in P&G's tech-innovation group. I clear the security and risk path so the company can adopt AI fast — without dropping the guardrails. I came up through cybersecurity (three promotions in three years), and I build AI products on my own time (AJ AI, receptr). So when a new tool lands on my desk, I've usually built one like it.
I work between innovation and risk at P&G: I clear the risk path — third-party risk, iRisk, governance — so scouted AI tools and startups can actually start POCs and scale. My foundation is cybersecurity (iRisk, third-party risk, threat and data protection), and I automate the work where I can. The InfoSec agents started as me trying to automate my own job — I couldn't run that data through a public tool, so I built it in Copilot Studio. I build AI products on my own time too, so I'm not guessing about the tech I review. The numbers below open if you want the proof.
The reason I can tell whether a tool is safe and whether it actually works: I build the same kind of thing. At work and on my own.
An AI receptionist for local businesses. It answers every missed call and text, 24/7, captures the lead, and follows up. Multi-tenant and deployed, handling real conversations.
An autonomous AI operator I built. One audited loop — classify → route → execute → persist — with a human on every irreversible call. SQLite is audit-only; a markdown vault is the source of truth. A doomloop detector and hard per-model budget caps stop a runaway agent before it spends. It exposes itself over MCP and routes across a local model plus Claude tiers — the same build-vs-buy tradeoffs my day job evaluates, at personal scale.
On my own initiative, I built a Copilot Studio assistant to automate my own InfoSec risk-clearance work — routine policy and review questions — saving roughly 10 hours a week. It proved the concept that became AskInfoSec.
P&G's company-wide multi-agent InfoSec system — a front-door triage agent plus domain agents for third-party risk, iRisk, policy, application compliance, and IT/OT. I contributed to it, building on the prototype I'd shipped.
A research engine that hunts for a real, defensible market edge. I backtested ~18 strategy families out-of-sample over 16 years, found no durable edge, and published the negative result. The edge is the control, not the signal. Paper-only.
// Completed an accelerated BSIT + MBA in four years. · Cyber@UC / OWASP · INTERalliance of Greater Cincinnati
I built an assistant on top of my own background. Ask it about my experience, my security work, or what I've shipped.
Grounded in AJ's real background. For anything it won't answer, reach out directly.
07 — Contact
Hiring, partnering, or comparing notes on security and enterprise AI — I'm easy to reach.