About
I'm currently a cyber lead / L3 tech at Propulsion Tech, a next-gen MSP helping organizations focus on advanced cybersecurity, compliance, and AI & Automation. Day to day that means managing the security stack, building MCP servers and the internal clients that consume them, detection pipelines, and red-team evaluations that turn frontier models into something a SOC can actually rely on. The client work stays inside Propulsion; what's public here is servers. Most of the work lives in the Microsoft stack — Defender XDR, Sentinel, Entra — because that's where the customers are. I also operate a full MSP security stack day to day: Huntress (EDR, ITDR, SAT), ThreatLocker for zero-trust app control, and a long tail of email-security, DLP, and RMM/PSA tools that I won't bore you with here.
Certifications
- Anthropic Claude Certified Architect – Foundations (CCAR-F), Sept 2026 — Credly
- Microsoft SC-200, Security Operations Analyst, Sept 2026
How I think about this work
The interesting thing about doing this work inside an MSP is that you can't hide behind theoretical threat models. Every detection has a customer attached. Every false positive is a real analyst getting paged at 2am. That constraint makes the AI-security questions sharper than they are in a vacuum — you can't ship a clever evaluation harness if it doesn't survive contact with a fifteen-tenant queue on a Monday morning.
How this site is built
Strict Content-Security-Policy. HSTS preload. No third-party trackers. The headers are the proof. I do not ship personal infrastructure that fails Mozilla Observatory.
Public credibility
- mcp-sentinel-detection-engine v0.1.0 — Sigma → KQL → Sentinel Terraform pipeline exposed as MCP tools to Claude. Read-only Azure auth via certificate-based OAuth. Shipped 2026-05-12.
- 1st place of 1,602 teams, Huntress CTF 2026: AI Arena (561 of 561 challenges cleared), with a writeup
The rest of the working history is on LinkedIn. The interesting parts are on GitHub.