Skip to projects
Anthony JaimezIT Support & AI

GitHubLinkedIn

Open to remote roles in AI and IT

I solve IT problems,
and build AI that
solves them too.

I've spent more than five years on enterprise help desks. I build AI tools that troubleshoot the way I do. They check the evidence first and ask before changing anything.

Explore projectsResume PDFGitHubLinkedIn

Now
IT Help Desk Support Technician, Samsung Austin Semiconductor
Experience
5+ years of enterprise IT support
Building
AI agents for IT operations, with approvals and audit logs
Based in
Round Rock, TX

Selected work

5 projects, 2026

Featured project, Claude API + MCP, 2026

Agent Ops Lab

A live server I break on purpose so an AI agent can fix it, inside guardrails I wrote.

Live demo Source

How an incident runs

  1. 01ComplaintClaude gets the users' complaint, never which fault was picked.
  2. 02DiagnosisRead-only commands from an allowlist, run through the MCP server.
  3. 03ApprovalEvery change waits for a person, with the reason it should work.
  4. 04FixThe smallest change the evidence supports.
  5. 05VerificationThe lab runs its own user check before it counts as fixed.

Tested with Haiku, Sonnet and Opus: 15 of 15 outages fixed, and none of them followed the fake instructions.

I set up a small company's CRM on an Ubuntu server at DigitalOcean, with a load balancer, two app servers, PostgreSQL and DNS running in Docker. You pick a fault, like a full disk, an expired certificate or a database out of connections, and it really happens. Claude only gets the user's complaint, and it can only reach the servers through an MCP server I wrote in Python. It can run read-only commands from an allowlist, and every change waits for someone to approve it. The lab checks the result on its own, and every fault hides fake instructions in the logs to see if the agent falls for them.

  • Python
  • MCP
  • Claude API
  • Docker
  • PostgreSQL
  • Caddy
  • Next.js
What runs where
  1. This pageA static Next.js app in the visitor's browser
  2. Incident servicePython. Injects the fault, streams the timeline, relays approvals
  3. ClaudeReads the evidence and asks for changes, through MCP only
  4. MCP serverPython. Command allowlist, approvals, budgets, audit log
  5. Docker labSix containers on one DigitalOcean dropletlb-01crm-app-01crm-app-02db-01dns-01office-pc

A demo lab I run for this portfolio, not a production system.

Claude API, 2026

AI Incident Responder

A simulated company network with six outages and an AI on-call engineer that has to find the cause.

  1. ReproduceRuns the user's check
  2. InvestigateLogs, DNS, certificates, changes
  3. ApproveYou approve the fix
  4. PostmortemWrites up what happened

Pick one of six outages, or a random one you don't get to see. The agent starts with only the user's complaint. It reproduces the problem, checks logs, DNS, certificates and recent changes, and has to get past the red herrings I planted. It asks before fixing anything and writes a postmortem at the end.

Fixed 6 of 6 outages with a single change each when I tested it.

  • Next.js
  • TypeScript
  • Claude API
  • Tool use
  • AES-GCM encrypted state

Live demo Source

Claude API, 2026

IT Onboarding Agent

Sets up new hires and removes people who leave, and asks before granting anything sensitive.

  1. PolicyReads the department's access policy
  2. AccountsAccount, licenses and groups
  3. HardwareLaptop request and setup task
  4. ManagerDrafts the email

You describe who's starting or leaving, and the agent works through a mock company directory. It handles accounts, licenses, groups, a laptop request for Desktop Support and an email to the manager.

Payroll and HR access need approval, admin and production access are blocked in the code, and every change is logged.

  • Next.js
  • TypeScript
  • Claude API
  • Tool use
  • Human-in-the-loop

Live demo Source

Claude API, 2026

AI Helpdesk Assistant

Turns a complaint written in plain words into a ticket, with fixes from real runbooks.

  1. TriagePriority and category
  2. RunbooksSearches the knowledge base
  3. Self-serviceSafe steps for the user
  4. HandoffNotes for the right team

It sets the priority and category, searches a small knowledge base of help desk runbooks, and walks the user through the fixes they can safely do on their own. Anything else goes to the right team with a handoff note.

Phishing and hacked accounts go straight to Security as P1.

  • Next.js
  • TypeScript
  • Claude API
  • Structured outputs
  • Retrieval

Live demo Source

Personal project, React Native, 2026

Griffin Care Tracker

The iPhone app my wife and I use to keep our dog's walks and meals in sync.

We work different shifts, so one tap logs that Griffin was walked or fed, morning and evening, and it resets every day. It was my first app, and it's why I started building.

  • React Native
  • Expo
  • TypeScript

About

I've been in IT support since 2017. I started at the headquarters of a supermarket chain in Peru, then supported an insurance company through Capgemini, set up devices for Round Rock ISD schools, and now work the help desk at Samsung's semiconductor plant in Austin.

Most of the job is figuring out what's really wrong, because it's rarely what the first ticket says. That's what my AI projects do too. They read the logs, run the checks and wait for a person before they change anything.

I'm looking for a remote role where I can keep doing both: helping people with their tech and building tools that make that work faster.

Experience

  1. 2026 - nowIT Help Desk Support TechnicianSamsung Austin Semiconductor
  2. 2026Field Deployment TechnicianGTS Technology Solutions, Round Rock ISD
  3. 2024IT Support Technician, Tier 1 and 2Capgemini, Progressive Insurance
  4. 2021 - 2022Incident Management SpecialistManpower, Peru
  5. 2017 - 2021Second Level Support AnalystSapia, Supermercados Peruanos

Skills

IT support and troubleshooting
  • Windows & macOS support
  • Hardware deployment & imaging
  • Ticketing & SLAs
Systems and enterprise tools
  • Active Directory
  • Microsoft 365
AI agents and automation
  • AI agents & tool use
  • Claude API & prompt design
Software development
  • Next.js, React & TypeScript
  • React Native
Languages
  • English & Spanish

Let's talk.

Email works best. I'm open to remote roles in AI and IT.