The lab, the agent system, and the habit of writing down my reasoning aren't a flex. They're the conditions that make contributing possible from where I'm starting. Cross-vendor testing only exists because the lab exists. Honest comments only exist because the written record holds me accountable to my own thinking. The recovery infrastructure exists because I broke things on the way here and am going to break things again.

A few examples of how this plays out, since process language without specifics is just process language:

• I bailed cleanly on a frida investigation last week when an internal audit showed the real time budget was 6 to 13 hours instead of the 1 or 2 I had estimated. The work was captured for future reference, not abandoned.
• I held scope on the AAudio PR when a maintainer offered to expand it to upstream PulseAudio. Merge risk on the already-approved PR wasn't worth the new ground. The expansion got its own future-workstream entry.
• I caught a four-year framing error in my own PipeWire prep plan before opening the workstream, by running a basic "what's already in the tree" check before doing the activity-signal research that had been the plan.

AI assistance is the tool I use not the controller. The choices, the on-device verification, and the corrections are mine. When I get something wrong (and I do), the correction is in the same thread.

The OSS work is one face of a larger thing. Actively building skills across the territory I'll need for the next chapter.

• Coursework completed: Anthropic AI Fluency, Google AI Essentials, 100+ hours of Coursera across AI/ML and adjacent topics
• Currently studying: CompTIA Security+ (cert prep, active), Linux, Python.
• Daily scripting practice in PowerShell, bash, and Python. AI-assisted, but the AI is the tutor as much as the tool. Every script I write, I ask why each piece works the way it does, then read the docs to confirm. Goal is to write scripts I can defend without consulting the AI afterward.
• Writing down the reasoning behind every non-trivial call I make is itself a learning practice. Forcing the articulation surfaces the gaps in my understanding before they ship.

I am not a developer. I am someone who is learning to do the work, using every tool that helps without pretending the help isn't there.

Two reasons.

One, I'm learning out loud. The notes I keep and the public comments I post are notes I write to my future self, but they're also notes anyone else can read. Someone going through the same path I went through (no CS degree but has the drive, passion, and mindset) might find the trail useful.

Two, I want to be wrong publicly so I can stop being wrong faster. The honest AI-disclosure pattern, the bail decisions, the "happy to be wrong" framings exist because someone catching me means I get to learn something. If you read something I posted that's mistaken, please tell me.

Current pulls: development and testing across the multi-OS lab, CI/CD pipelines, and the overlap between networking and security. The networking-security one gets most interesting when LLM assistance is the thing running between them. If any of that lines up with contributing to an indie project or joining a team where the lab habits actually pay off, the email's up top. Career pivot into tech is the long arc. Soaking up knowledge and wisdom is the route. The work itself is most of the fun.