From Navy Fiber Optics to Python: A Career Pivot Into AI Development, In Progress
Houston, TX · a field-notes look at an IT career in transition
This one's more personal than most posts here. I've spent close to two decades around IT — military electronics, enterprise help desks, Active Directory, PowerShell — and I'm now in the middle of the harder pivot: from IT support into AI development and Python programming. Not the polished "I made it" version of this story. The actual in-progress version, certifications half-stacked, still driving for delivery apps to keep the lights on while I study.
The Foundation: Fiber, Satellites, and a Navy Base in Hawaii
The starting point was Pearl Harbor, working as an Aegis Computer Networking Technician for the U.S. Navy — repairing military computer equipment that ran the gamut from fiber optics to satellite dishes to high-voltage power supplies. It's the kind of work that doesn't show up directly on a "Skills" section of a resume, but it's where the habit of methodically diagnosing a system you didn't build, under real time pressure, actually gets trained into you.
A Decade of Building IT Support Chops
What followed was a string of end-user support and technical analyst roles across some genuinely different environments — an energy major, a hospital system, a healthcare distributor, a benefits administrator — each one adding a layer:
The connective tissue across all of it: automation and scripting kept showing up as the part of the job that actually held my attention — writing a PowerShell script to handle driver installs instead of doing them by hand, batch files to move data with RoboCopy instead of dragging folders one at a time. In hindsight, that was the tell.
School While Working
The degree work happened in parallel with all of the above, not before it — a Bachelor of Science in Information Technology from Western Governors University (3.0 GPA, completed September 2019), an Associate of Science in Information Technology from Sanford Brown (3.9 GPA), and a stretch of undergraduate coursework at the University of Houston that included Calculus III, Partial Differential Equations, and Numerical Methods alongside Object-Oriented Programming in C++. None of that math was required for help-desk work. It's there anyway, and it's part of what's making the current pivot toward AI feel less like starting over and more like circling back to something.
The Pivot Starts Now
Two certifications earned in early 2026 mark where the shift actually began in earnest:
Worth being honest about what these actually are: AI-900 is a conceptual, no-code fundamentals exam — it proves you understand machine learning, computer vision, NLP, and generative AI concepts on Azure, not that you can build production AI systems. PCEP is the entry rung of Python certification, not a substitute for building real projects. Neither one gets you hired as an AI engineer by itself. What they're actually good for is exactly what they're being used for here: a structured, credible signal that the pivot is real and already underway, and a foundation the next certifications build on.
A Necessary Correction: AI-102 Just Retired
Here's something worth flagging plainly, because it changes the plan: the next step on this path was going to be AI-102 (Azure AI Engineer Associate) — but Microsoft retired that exam on June 30, 2026. It's gone; there's no sitting for it anymore. Its replacement is AI-103: Developing AI Apps and Agents on Azure, leading to the new Azure AI Apps and Agents Developer Associate credential. The shift isn't cosmetic — AI-102 tested integrating pre-built Azure AI services, while AI-103 is built around generative AI, Microsoft Foundry, and agentic architectures, which is a more accurate map of where AI engineering work actually is in mid-2026 anyway.
Redirect study time from AI-102 material to AI-103 directly, rather than chasing a retired exam. It's more work up front, but it's a certification that will still mean something in two years instead of one that stopped being earnable this summer.
What the Grind Actually Looks Like Right Now
The part that doesn't usually make it into a LinkedIn post: this transition is being funded by delivery driving and gig work — Papa John's, Uber, Lyft, Instacart, Postmates — running alongside contract tech support work. That's not a footnote to hide; it's the actual current state of a mid-career pivot happening without a training program or a severance package behind it. Studying Python and Azure AI concepts between delivery runs is slower than a full-time bootcamp, and it's also exactly how a lot of real career changes get made — in the margins, on a schedule that has to work around rent.
The Skills That Actually Transfer
The automation instinct built over a decade of IT support turns out to be a real bridge, not just a nice narrative:
- PowerShell and batch scripting for driver installs and user management maps directly onto Python scripting for data pipelines and task automation.
- RoboCopy-based data migration is, structurally, the same problem as building an ETL step in a machine learning pipeline — move data reliably, verify it landed correctly, handle the edge cases.
- Diagnosing legacy systems under time pressure — Citrix, Active Directory, SCCM — builds the same debugging patience that shows up constantly in model development and data wrangling.
- Documentation habits from help-desk ticketing translate directly into the kind of clear technical writing AI/ML roles actually need for model cards, pipeline docs, and handoffs.
No AI development job yet. Certifications in progress, not complete. Still doing gig work to cover the gap. This post is a marker for where the pivot is right now, not a highlight reel of where it ends up — the plan is to keep updating it as the AI-103 study progresses and the first real automation or ML projects get built and shared.
What's Next
- Redirect Azure AI study from the retired AI-102 to AI-103 (Developing AI Apps and Agents on Azure)
- Continue toward the NVIDIA-aligned Associate Generative AI (NCA-GENL) curriculum already underway
- Move from certification study into small, shareable Python automation projects — the kind that turn "PCEP certified" into an actual portfolio
- Keep documenting the process here rather than waiting for a finished narrative to post about
If any of this overlaps with your own path — Navy or otherwise, IT support or otherwise — feel free to connect:
No comments:
Post a Comment