This last week I spent the easiest $200 on “art” (read: paperweight) that I ever have. I bought a formerly $3,000 “Trash Can” Mac Pro. This token object is a physical representation of an early inflection point in my career where I made the transition into the world of DevOps.

I started my career as a mobile developer. Despite some early success [1], it was not a good fit for me. I've never been a good graphic design person or really artsy in any way. As a kid, when I tried to cut dinosaurs out of coloring books to color in the lines, I cut a full centimeter-ish around the outline instead of trying to finesse the scissors to smoothly cut as close to the line as possible. An extra bit of paper around the border was good enough for me. This type of attitude towards aesthetics did not translate well to trying to create a pixel perfect implementation of a design.

I was given an opportunity to move from being an iOS developer to an SDET. I did really well at writing functional automated tests for our iOS app, particularly ones that involved complex integrations. I found myself really enjoying gluing systems together that weren’t necessarily meant to be put together in an automated fashion, but still figuring out a way to make it work.

The question eventually became: how do we orchestrate these automated tests? When do we run them? How do we run them? That led to Mobile DevOps. We had to figure out how to run a bunch of tests in parallel for PRs and main branches on Mac Pros.

For those who aren't aware, Apple discontinued their enterprise-grade Xserve product line in 2011. This is a problem because enterprises with apps (which is few of them in 2011 to “all of them” when I started doing DevOps in ~2015), still needed a way to reliably build, test, and deploy iOS apps to the App Store. Most people got by with a Mac Mini or two sitting under their desk or in a storage closet.

Anyone with a team of more than five people quickly needed more Mac hardware to scale. Enterprises needed to ensure that any changes merged in to the main branch did not break anything in their app using their primordial mobile CI/CD pipeline. You officially cannot build iOS apps on non-Mac hardware [2]. If you wanted to do things on the up and up, you had to eventually get a rack or two of consumer-grade Mac Pro hardware in a data center.

Don't let the name fool you. Mac “Pro” is a misnomer. It may have had higher specs than the Mac Mini, but it was still very much just a prosumer computer. It was not designed to run 24/7 in a data center. We had to do a lot of custom configuration and hit a lot of “gotchas” to ensure that the Mac Pros ran reliably. Every computer that Apple made after Xserve was shuttered assumed that there's only going to be one or two users who are going to be using the computer, one at a time.

This is a problem in a Mobile DevOps pipeline where we run dozens of builds on one computer all day, every day. Eventually, file system cruft accumulates, memory leaks, and you resolve a lot of these issues with a good ol’ clean-up job and full restart once a day.

It was a big deal when we could eventually virtualize these Mac Pros. This simplified our MacOS and Xcode upgrade process because we could just make a golden image that was validated against our pipeline and then clone that around to all of our different Mac Pro hosts. If we had an issue with a VM that we couldn't debug within a couple of minutes, we just deleted the VM and then cloned a new one. It was really nice to finally treat our MacOS systems more like cattle than pets.

Glad you asked. I did not go to school for DevOps. Mobile DevOps, even. It's such a weird little corner of the software world that there's only a handful of companies in the space, and it certainly did not exist when I went to college in 2009.

I thought I was going to be a mobile app developer. After all, I got featured in Forbes online [1]. I felt like I really made it, like I was almost in a Bruno Mars song.

But then I had to re-skill. Thankfully, I was able to learn this new skill on the job.

It took a lot of time and effort. I didn't even mention how I had to learn Jenkins at the same time (collective groans from the audience). I had to learn how Jenkins works through all of its iterations: freestyle jobs, to scripted pipeline, to declarative pipeline.

I can probably set up a basic mobile DevOps pipeline today if I wanted to, given what I remember about the process and “gotchas” that are now several years out of date, but now I'm augmented by AI Agents.

I have mastery of the process, but not necessarily of the details. This is me cutting out dinosaurs out of coloring books all over again.

A lot of people aren't happy about having to re-skill, given how much time, effort, energy, blood, sweat, and tears they have put into refining their programming skills over potentially a lifetime. Some people really enjoy cutting close to the dinosaur outline and get a thrill out of how smooth and unfrayed they can make the paper as they cut it. They are real craftspeople who feel like they are losing their identity.

I had the same feeling too recently of losing part of my identity. Last year in 2025, after a number of injury setbacks, I decided to stop competing in Jiu-Jitsu. I put too many miles on my body. I no longer could really see the fulfillment ROI of competing anymore. I felt like I went and accomplished the spirit of the thing that I needed to do in 2022 ADCC West Coast Trials.

I proved myself to be somewhere on the absolute Jiu-Jitsu skill scale rather than a relative local or gym-level scale. My life would not significantly change if somehow, I did in fact win ADCC Trials at this point. I would probably keep my same job and my same insurance.

It still didn't feel good to basically be forcibly retired from competition. I had to really question the “why” behind what I was doing Jiu-Jitsu for anymore. I was a Brown Belt instructor, had been training for 12 years, who lost the primary flame that kept him on the mats every day. If I could no longer strive to be the best competitor, what was I doing this for? I had to find a new “why”.

I think that's how I can relate to software craftspeople who feel a loss of identity when their skills that they have worked so hard on are no longer directly relevant. It's hard to be positive about a technology that directly took away your source of self-esteem, mastery, and potentially income. Same with me and Jiu-Jitsu.

I have a very different relationship with Jiu-Jitsu now than I did when I was competing. As I was coming up, I couldn't train enough. I could not get enough of it. I was training with the dial on 11. I was willing to take on any and all challengers, and challenge anyone. I was fearless and free.

Now I'm turning the dial back to a bare minimum drip of training 2-3x per week to minimize potential risk for injury, while still enjoying practicing the sport with trusted training partners.

My “why” now is not centered around my individual performance of Jiu-Jitsu, but rather my deeper understanding of how I can communicate it to others. How I understand things in Jiu-Jitsu and how I communicate them happen to be informed by something called the “Constraints-Led Approach”, or CLA for short.

Coaching with CLA feels suspiciously like Agentic Vibe Coding, and I got a two-year head start on it. I’ve been coaching people with CLA by defining boundaries, constraints, “guardrails”, invariants, and goals for my students to focus on. They interact primarily in a live “mini-game” fashion, pressure testing everything that they do against a resisting opponent who has their own goals for each “game”.

My why for Jiu-Jitsu is to encourage others to get whatever they want out of Jiu-Jitsu, whether that be just a fun way to exercise, self defense, or to become a world champion like I once dreamt of. All are very valid goals. All can train in the same room with each other. It was deeply meaningful and fulfilling for me to pursue my dream, so I want to enable others to do the same. I lived that part of my life fully with no regrets. It would have eaten at me otherwise with “what ifs”.

I think software craftspeople have to go through their own sense of turning this loss around and transmuting it into something positive. Glass half full versus half empty. I know this is easy to say and hard to do. Ask me how I know. But I feel like we will get much better results as an industry as more people get bought into this new way of coding.

We will go through certain growing pains like a single nine of reliability (~90% uptime) from Claude in March 2026. But the growing pains mean that lessons are being learned. We're still figuring out how to use this brand new Agentic coding technology that was airdropped on us about a year ago.

As more companies begin to adopt this new way of coding, we'll see more single nines and continued smug “I told you so’s” from craftspeople who want to hold on to the old days. We're not going to “revert” this industry-wide “code change” during this “incident”. This is a fix-forward situation.

How we fix forward, nobody really knows yet. We're all still figuring it out. But it will help if we had more people on board. More craftspeople who know what's going on and can give us better insights and potential processes to implement moving forward.

This is how I felt going from mobile developer to Mobile DevOps, an incredibly niche space where very few people had it all figured out. I never took a class on or had any experience in Mobile DevOps before I started working in it, but there were still some fundamental things that I was able to carry forward and use as scaffolding in this new world.

Let's figure it out together.

If you haven't already, you can join our Discord to connect with other like-minded AI-enabled software engineers and discuss the latest trends, tips, tricks, and techniques.

It's a great place to further explore topics that we discuss on the live stream and where I put all of the links that we pull up every stream.

[1] I'm going to take this opportunity to flex that I was featured in Forbes online because I knew enough Android to create a proof of concept spyware for Google Glass in our graduate-level security class in college. We got Forbes (and Google’s) attention because our professor flexed on Twitter. Taste and distribution struck early.

[2] Apple is notorious for being at the minimum a pain to deal with, to being actively hostile to developers who are paying Apple $99 a year to build on their platform.