IT Admin

About the role The IT Administrator owns day-to-day information technology, network and cloud infrastructure, business-systems support and development, and cybersecurity acrossSkyetonInc.’s U.S. locations. This hands-on role supports employees and operational systems; administers endpoints, identity, networks, servers, collaboration platforms, source control, and cloud services; andmaintainsreliable, secure access to the technology the business depends on. The position also specifies, reviews, tests, and deploys changes to internal business systems and integrations, including AI-assisted development completed under human engineering review and established software-governance controls. The IT Administratoris responsible forsecurity hardening, monitoring, incident response, access control, documentation, and technical compliance evidence supporting ITAR/EAR, Controlled Unclassified Information, NIST SP 800-171, DFARS View phone number on click.appcast.io, CMMC, and customer requirements. The role is an individual contributor initially and may grow into leadership of a small IT and security team. What you’ll do End-User It And Endpoint Management Own the full endpoint lifecycle for everySkyetonInc. US employee:procure, image, configure, deploy, patch, encrypt, support, recover, and retire laptops, desktops, mobile devices, and peripherals against a documented standard build. Provide responsive Tier 1 and Tier 2 support across all sites for endpoints, business applications, printing, conference-room technology, and shop-floor systems. Be the person people are glad to call. Administer the corporate email, collaboration, file-sharing, and cloud identity platform: directory and group management, multifactor authentication, conditional access, licensing, and permissions. Run onboarding and offboarding end to end with HR: account provisioning, group and role assignment, equipment issue, and — the part thatactually mattersfor security — same-day, verifiable access removal on departure. Maintainan accurateasset, license, warranty, and configuration inventory.Trackspend and lifecycle so equipment refresh is planned, not reactive. Network And Infrastructure Own the wired and wireless networks at everySkyetonInc. site: switching, VLAN segmentation, firewalls, wireless, structured cabling, internet and carrier services, and theremote-accesslayer. Segment the production floor from the office. Network-connected manufacturing equipment — CNC, cure-room controls, test stands, booths, inspection gear — belongs on its own segment, with deliberate and documented rules about what may cross. Stand up the network and IT infrastructure for new sites and facility build-outs, from the cabling drawing to the day the first person plugs in. Administer Linux and Windows servers and the company’s cloud footprint, in coordination with the existing infrastructure-as-code and following it — anything provisioned by hand and never mirrored into code is a rebuild failure waiting to happen. Own the remote-access and zero-trust network layer: device enrolment, group membership, access policy, and the identitychecksbehind it. Own backup, restore, business continuity, and disaster recovery. Test the restores and document the results — an untested backup is a hope, not a control. Monitor availability, capacity, storage, and alerting; investigate and resolve before users report the problem. Business-Systems Development Specify, direct, review, and ship changes toSkyetonInc.’s business systems: the ERP (Python / JavaScript) and its custom document types and extensions; the leadership dashboards; the accounting, card-spend, and HR integrations; other internal API integrations; and the supporting infrastructure-as-code. Develop with AI coding agents — and own what they produce. Most of the code in these systems is agent-written under human direction and human review. That is how acompanythis size moves at the speed ithas to, and it only works because a competent engineer stands between the agent and production. Concretely: Specify before you generate. What comes back is only as good as the problem statement, the stated source of truth for every value, and the constraints you handed over. Vague prompt, plausible garbage. Review as though you wroteit, becauseyou are answerable for it. Read every line. Check it against the design, the data model, and the source-of-truth doctrine — not just against “does it run.” An agent’s output is a proposal, never an authority. Hunt the failure modes agents actually have: a value copied instead of pulled from its system of record; a silent fallback that swallows an error; a plausible-looking match that attaches the wrong record; an assumption that quietly disagrees with the rest of the system. These do not look like bugs. They look like working code. Test what youapprove. Agent-written code gets the same tests, the same gates, and the same staging rehearsal as anything else — more, if anything, because nobody’s fingers walked through the logic line by line. Know when not to use one. Some work —a securitycontrol,a credentialpath,a schemamigration against live data — is worth doing slowly and by hand. Improve the harness itself: the project rules, the prompts, the tests, and the CI gates that make agent output reliable. When an agent makes the same mistake twice, that is a missing check, not bad luck. Work the governance pipeline as written, every time. Our software development plan is short, real, and enforced. In practice that means: A work item first. Nothing gets built before there is a work item defining it. A plan document, not a comment. The planstatesthe problem, the source of truth for every value, the design, the companion merge requests, verification, rollout, rollback, what “done” looks like, and the open questions. It is agreed beforecodeis written. A merge request with a green pipeline. Fast-forward-only, rebased, squashed. A gate that cannot fail does not exist — no advisory checks, and nothing merges through a known-red pipeline. Rehearse on staging before production, sametooland same flags, from a fresh production restore. First-time code paths neverexecute onproduction first. Productiondeploysare deliberate and human-run — a prepared release script that a human reads and runs, never a CI button and neverbecausea merge request landed. Every ERP extension is registered in the extensions register in the same commit, so the next person can trace every departure from stock. Done means finished. A change that needs something outside itself is not mergeable, and work does notendwith a list of loose follow-ups. The rest of the development scope: Respect the source-of-truth doctrine. Data lives only in its system of record and is pulled from there. Nothing copies, caches, or re-keys data as a workflow. Before you write anything that holds a value, you ask where its source of truth is and whether what you are building is a view of it or a fork of it. Forks are theexpensive kind ofmistakeand they are rarely obvious at review time. Support and administer the systems you build: ERP user and role provisioning, workflow and approval configuration, print formats, scheduled jobs, integrations, upgrades, and the un #J-18808-Ljbffr Skyeton