A personal reference
Jackagonz
AI implementation. Security and infrastructure. The spread is the point.
I build the layer between a capable model and a system people can actually use.
My background is in security and infrastructure. I came into software laterally and learned it through the systems I needed to build. That put me at the point where models meet context, tools, permissions, evaluation, and deployment.
This site is wider than that work. I'm an intensive polymath by temperament. Security, infrastructure, software, AI, systems design, cars, real estate, and the philosophy of craft all sharpen the same habit: take the thing apart, learn what controls the outcome, and build until it holds up.
Breadth without depth is dabbling. Breadth with depth compounds.
1Background
Lately I've been deep on agentic AI and the Model Context Protocol, the tooling that lets AI assistants call real tools and take real actions. I look at it the way I'd look at any new attack surface. Inventory it, sort it by trust zone, grant only the access a task actually needs. I don't have settled answers on where any of it is going. I find that more interesting than pretending I do.
Systems should fail loudly instead of quietly. Automation should remove a step, not hide it. Documentation nobody else can follow isn't documentation. Most problems are smaller once you write them down.
Things I believe, at least this week
2Career
-
June 2025 to June 2026
The Privacy Office
Principal Information Security Consultant
Prior independent practice. Risk assessments, control design, policies and runbooks.
-
2024 to 2025
All Connected
Security Engineer
Worked across SOC/NOC operations, identity and infrastructure. Handled escalations and wrote repeatable procedures.
-
2024
Alexander Auto
IT Technician, internship
Ran Active Directory for the dealership and restructured access to least privilege. Killed a site-wide VoIP problem at the router.
-
2022 to 2024
Freelance
Independent IT Technician
IT support for households and small businesses. Malware and ransomware cleanup, and backups that actually work.
3Selected work
What should an agent be allowed to reach?
I reviewed my own MCP tooling by capability and data destination, then designed a more restrictive access model. The case distinguishes the design from deployed protection.
Finding“Available” and “active” are not the same word.
Green checks. An incomplete inventory.
A discovery gap survived the local checks because the implementation and verification shared it. The repair started with the population the workflow was supposed to cover.
A personal site with fewer moving parts.
Static content, local assets and a separate editorial workspace. The implementation is open to inspection, including the limits of its privacy and testing claims.
4Approach to AI
I use AI tools in my day-to-day work and in building this site. Disclosure should be the default, not something you have to ask about.
How I use it
- As a professional focus, not just a tool. Part of my current work is inventorying and securing the agentic AI and MCP tooling I use. I treat it the way I'd treat any new system with access to my accounts and data.
- As a drafting and automation aid: outlining, first drafts, and repetitive scripting work. Everything published under my name, this site included, is reviewed and edited by me before it goes live. AI-assisted is not the same as AI-authored-and-unchecked.
- As support for research and organization, not a substitute for responsibility. Consequential recommendations remain mine to verify and defend.
The portrait
This is an AI-generated portrait based on my appearance, not a photograph. I use it to avoid publishing another high-resolution photo of my face.
That does not make the image impossible to copy, identify, match or misuse. The disclosure describes what it is, not a guarantee of anonymity.
The career history and earned certifications are real. The degree remains in progress.
What this means for you
- Nothing on this site is AI-generated content published without review.
- The portrait is the one generated image here, and it's labeled as such.
- If you're a recruiter or collaborator and want specifics on how I use AI in a given context, ask. I'll give you a straight answer.
5Interests
- Paper notebooks
- Moleskine for years, and lately I've been trying to find something that beats it. Numbered pages, lay-flat binding, A4 or bigger. It's a weirdly specific hunt and I'm enjoying it.
- Games, both kinds
- Stardew Valley and Arc Raiders on a screen. Apocalypse World around a table. The tabletop stuff scratches a different itch, closer to writing than to playing.
- How games are built
- Game Maker's Toolkit, solo-dev logs, anything where somebody explains why a mechanic works instead of just showing it.
- Film
- A24 puts out most of what I like. Shutter Island is the one I keep going back to.
- Privacy and the surveillance surface
- Adjacent to the job, but I'd care anyway. What gets collected, who ends up holding it, and how little of that most people ever agreed to.
- Workshops and desks
- I have a long-running plan for a build with more drawers than anyone needs. Not started. Fully designed in my head.
- Plants with a shape to them
- Silhouette over flowers. I want something architectural in the corner of a room, not a bouquet.
- Cooking, mostly meal prep
- Batch it Sunday, eat well all week. Same instinct as the automation stuff, honestly.
- Cars and real estate
- Two long-running interests I read about far more than I currently act on.
6Education and credentials
- Education
- B.S. Cybersecurity and Information Assurance, Western Governors University. Degree in progress; expected 2026.
- Certifications
- CompTIA PenTest+ Security+ Network+ A+ Project+
- Availability
- Open to remote. Southern California based.
7Contact
Roles, consulting work, or a straight answer about anything here. Hit me up as you'd like.