Log in
| 1 | ||
| 1 | ||
| 1 | ||
| 1 |
(a rewrite by ChatGPT)
For some reason, my government wasn’t verifying that I was a real person when I tried to register an LLC. This process was supposed to be instant. Previously, I mailed a physical form to get it started, and I’ve been checking name availability obsessively, hoping to see some confirmation.
Today, I followed up with a phone call. They found no record of my submission—hmmmm. The postal service should have delivered it by now, unless it got rerouted internally. There was even a money order in that envelope! Fortunately, a kind representative on the phone pointed me to the correct online system to escalate manual verification that I am indeed human.
I’ve also been busy with the local tech networking organization, where I recently joined the board. Our first meeting of the year is postponed until mid-February, so we’re taking it slow. The initial series will be led by a fantastic colleague, with different board members rotating each week. I’m up first, presenting a simple coding example of breadth-first search—covering roughly the first two chapters of our book.
Once her series wraps up, it’s time for my series, titled Building with Huggingface. There’s a senior machine learning engineer—an enthusiastic professional from a firm trying to use AI as a router for their accounting suite—who wants to dive into an in-depth exploration of the Huggingface Transformers library. However, my approach is different:
I want to show how AI can transform you from a mere end user into an empowered developer—harnessing powerful, research-backed tools to build real products.
The final part of my day was, unfortunately, the least productive. There are technologies I’ve long avoided because of time constraints, but now that I have some downtime, I decided to tackle them—starting with DevOps and, specifically, Kubernetes.
I still don’t understand why setting this up has always been so challenging. Today, I attempted to use the supposedly easy setup option, minikube. My Artix Linux installation isn’t playing nice with virtualization (even though I’ve had Docker running on this very machine before). Now, Docker, Runpod, and VirtualBox all refuse to cooperate.
I even rented a CPU runpod instance to set up the prerequisites. Despite renting a box with 8 CPUs, it complained about having fewer than 2 CPUs. There was a --force option to bypass the check, but Docker eventually failed while pulling images. I can’t fathom why some setups are so persistently problematic.
Some might say I’m inexperienced, but I’ve deployed countless systems, built Linux from scratch (beyond LFS), and compiled massive dependency trees. Clearly, I’m no novice. After years of following official instructions on even the most standard Linux distributions and hardware, it seems their software isn’t built for the average machine.
Perhaps that’s why tools like Kubernetes and Docker exist: to support software that demands a level of complexity in deployment. It’s like this:
Nobody:
"Our software doesn’t need to be broadly deployable."Kubernetes:
"That’s what DevOps tools are for."
If everything were deployed on Kubernetes, many of these installation issues might never arise.
Honestly, this is incredibly frustrating. Their software is simply inadequate.
Every challenge—whether it’s navigating bureaucratic hurdles, pioneering tech education, or wrestling with DevOps—offers its own set of lessons. The journey is full of surprises, and I’m here to share every step.
Just like the image says. The most common reason for dev-ops failures is dev-ops.
Stay tuned for more adventures in tech!
Here is the original before chatgpt touched it: https://goatmatrix.net/c/DevBlog/8YY64GbshT
Notice in this version the phrase kind representative that chatgpt made sure to put in bold, verses the prior version.
They seem about the same length. Both semi-interesting to read, but very dry and abstract. I'm afraid I've already forgotten most of what I might have thought I learned. It would likely have more impact if I could relate more, especially if I had practical experience in dev-ops and/or A.I.
They should have almost identical content. What I really want is chatgpt to just do formatting. But it tends to reword things as well. In theory it does make my writing more professional. Or it would if the sentences coming out of it always made sense. They don't.