570.

That’s the floor. The confirmed minimum of engineers who got walked out in March, badges deactivated, and severance package PDFs waiting on the other side of an HR Zoom call. This number is the floor, because two of the five companies that cut this month decided the headcount was nobody’s business. Verint Systems said “several hundred.” Digg said “a sizable portion” and then pulled their app from the App Store before anyone could ask follow-up questions. Strip out the acquisition restructures, and an air taxi company that discovered physics doesn’t negotiate, and what you’re left with is 74 engineers who were developers, IT, and R&D who walked in on a Friday and walked out with a box. That’s your number. The rest is noise, wearing a press release. [Source: layoffs.fyi]

WHAT’S MOVING

Labor

You Can Have the Job. Just Come Into the Office. The LinkedIn numbers this week tell a story the job board newsletters will not. DevOps Engineer postings: 6,418 total, 3,528 new this week. Healthy volume on paper. But run the location filter, and the market shows its teeth: remote positions are 902, hybrid plus onsite combined (“Hybrid” is just another way of saying “On Site”) is 2,608. Ratio: 0.35. For every posting offering genuine location flexibility, nearly three are telling you where to sit and when to be there. SRE roles are worse at 0.26. Platform Engineer roles sit at 0.30. Cloud Engineers at 0.31. The RTO argument is over. The question you should be asking is whether the salary bands reflect the commute they just added back to your life. Based on this week's sample data, the answer is: sometimes, and only if you negotiate. [Source: LinkedIn Jobs]

Technology

The Bots Took the Tag. A script I built to track what DevOps engineers are actually excited about on GitHub returned ten results this week. Nine of them were AI agent frameworks wearing a DevOps tag and keywords like a visitor badge they found on the floor, stuck on their shirts to complete their barely thought-out disguise. The one repo that wasn’t an AI framework was project-nomad, an offline-first survival computer built for zero-connectivity environments, which nearly doubled its lifetime stars in seven days. Engineers are not excited about DevOps tooling right now. They are excited about building things that work when the network doesn’t. That is a signal worth sitting with. [Source: GitHub Topics Search executed on 3/16/2026-3/22/2026 via API]

Vibe

The Quiet Collapse of an Open-Source Project: Deno laid off staff this week without a public announcement. Developers pieced it together from LinkedIn departure posts and a Reddit thread. A blogger wrote a pointed postmortem titled “404 Deno CEO not found.” It hit 204 points and 138 comments on HN. The community largely defended Ryan Dahl (the Node.js creator), a genuinely rare engineering talent while the comments ran hot on a deeper question: is there a single VC-backed open-source project that has stayed genuinely healthy over the long term? The room couldn’t name one cleanly. Bun got acquired by Anthropic. Deno got quiet layoffs and a mean blog post. Your readers build on this infrastructure. [Source: Hacker News]

THE REPO WORTH WATCHING

Crosstalk-Solutions/project-nomad - 7,178 stars this week, 12,933 total

A self-contained, offline survival computer. AI, critical tools, and knowledge designed to work anywhere, with or without connectivity, and with or without a cloud provider standing between you and your data. It nearly doubled its lifetime stars in seven days, which, given where the vibe is at this week, and in general, tracks. This is the same week the “devops” GitHub tag got colonized by AI agents; the top-velocity repo in the infrastructure space is about building things that work when the network is gone. Engineers are quietly bringing all the useful things on the internet back within the castle walls, or at the very least, planning on doing so.

VERIFIED VS. VIBES

“You Don’t Even Need DevOps” They Said, From a Vercel Blog

The Vibe: Vercel published a piece this week titled “Two Startups at Global Scale Without DevOps.” The argument: managed platforms have matured to the point where dedicated DevOps functions are optional overhead for startups moving fast.

The Verification: LinkedIn shows 6,418 active DevOps Engineer postings this week, 3,528 of them added in the last seven days alone. Whether or not companies intend to fill these roles that they actually posted remains to be seen, but the numbers alone tell a story. The federal government’s official labor market data (BLS JOLTS) is running behind because the government is shut down, which means the statistical floor under all of these narratives is currently a hole in the ground until fresh numbers drop, so until then we will have to rely on open sources like LinkedIn to get these numbers and see where they go week to week.

The Read: Vercel sells managed hosting. A world where startups don’t hire DevOps engineers is a world where startups buy more Vercel. Read the incentive structure before you read the headline. The argument isn’t wrong for every company at every stage, but it is published by the company that profits most from you believing it. Six thousand active postings say the market hasn’t agreed yet.

TL;DR - The numbers imply that this is just a hot take by Vercal’s marketing team, and not an industry reality.

THE QUIET THING

We were somewhere outside of the old internet as people my age remember it, that wide-open, beautiful and chaotic disaster that used to feel like it was built for actual human beings, when things began to change in a meaningful way. What once was has become a distorted facsimile of itself.

The people who inherited the beautiful chaos of the old internet began to see human behavior on that medium as sterile metrics. Engagement numbers, optimization loops, quarterly growth targets, and all the data about who is doing what turned every platform that was ever worth a damn into something that needed to be worth more. The gospel of platform growth treated actual humans like they were in the way, or worse, like they were the product. It stopped being for the people.

Digg died the week of March 13th. Again. Kevin Rose bought back the platform he started in 2004 and rebuilt the whole thing from scratch to be what it once was, putting a team on the field with real moderation and verified humans with the goal of creating a genuine place to put things worth sharing. Not long after the public beta launched, the bots found it within hours. Before a single real community had a chance to form, the automated accounts were already camped out like they were waiting for easy spawn kills. Voting on things no one had read. Building the appearance of a conversation that was not actually happening. CEO Justin Mezzell wrote the postmortem himself, which is the kind of thing a person does when they still believe honesty matters: “the internet is now populated, in meaningful part, by sophisticated AI agents.” The app has since been pulled from the app stores. The only content on Digg at the time of this writing is the eulogy.

This past week, I ran a script to find what DevOps engineers were actually excited about on GitHub as part of the vibe check for this issue. The script returns the top ten trending repositories for a given set of categories related to DevOps and Platform Engineering. Nine of those ten results were AI agent frameworks wearing a DevOps tag like a costume from a dead man’s closet.

The same week, a YC-funded company launched a product to help brands get cited by AI search engines called Sitefire. The machines are now writing content to be read by other machines to be summarized for humans who will never see the original. Somewhere in that chain the human got optional.

I have been looking for the original dream of the internet my whole career and I think I know what it was now. Not the retirement package, not the IPO, not the VP title. The dream the original builders were chasing when they stayed up until 3AM in dorm rooms and garages and server closets held together with cables some other engineer forgot were plugged in was a network that amplified human connection rather than replacing it. A place where distance did not matter and your idea was as good as anyone else’s idea and the conversation was real because the people having it breathed the air and walked on their own legs.

We did not lose that dream. We lost the thread of it somewhere around the third round of VC funding, when the metrics stopped measuring connection and started measuring synthetic values that exist in service of making the numbers look impressive in a deck.

But here is the thing about dreams: they do not die when you stop finding them. They die when you stop looking.

Someone is going to build the more human internet, and they will do so carefully, with intention, in service of something actually worth building. Someone is going to figure out how to put the humans back in the center of the network. Someone is going to solve the bot problem that killed Digg’s relaunch, get the community there before the machines do, and build the thing that makes human presence visible and meaningful again.

That someone is learning how systems work right now. In a classroom, on a VPS, on old server hardware rescued from an e-waste pile. That someone is going to find a bunch of other people like them and set off on some harebrained, underestimated quest with a beautiful ramshackle idea that will remind everyone that we are still here.

The dream of the old internet is not dead. It is waiting for the engineers who are young and curious enough to still believe it is possible, and skilled enough to actually build it. I know those people exist because I have met them, and they have given me hope.

What I saw crawling around the innards of tech this week may look like a warning. I see it as a critical inflection point, a moment where the current AI runup is showing its age and the tools it produced are finally mature enough to build something genuinely worthwhile with. The warning and the invitation are the same thing. It depends entirely on who picks it up.

FOR THE PIPELINE

The remote ratio data from this week’s job search tells a story every student in a tech program needs to hear before they build their job search strategy around remote-first assumptions. In the current DevOps, Platform, and Cloud Engineering market, on-site and hybrid postings outnumber remote postings nearly 3-to-1. That does not mean remote is completely dead. It does, however, mean geographic flexibility is now a negotiating asset, not a baseline expectation. Before you take an interview, research the company’s remote history, not just their current posting. A company that was remote-first in 2022 and is now “hybrid” is a different negotiation than one that was never remote. Know the difference. The LinkedIn Company Pages “Life” tab and employee reviews filtered by “remote work” will tell you what the policy actually looks like from the inside. That is free research that most candidates skip. Don’t skip it.

It’s about 6:42pm. We’re on the bus headed back to my place. There’s this loud fellow that embodies “TikTok Conspiracy Bro” having a loud conversation with a fellow I colloquially refer to as “Bus Bob,” sitting across the aisle. Bus Bob is always on the 6:35 and is usually quiet, so watching him engage with this loudly gesticulating interloper is a curious sight.

Conspiracy Bro had the look of a man who had, at some point in the recent or distant past, looked at conventional society and simply walked away from the negotiating table. Done with it all. The “social contract,” the general agreement we’ve all made to pretend we care what strangers think of us, he just wasn’t having it. His entire physical presentation radiated the serene confidence of someone who “I DiD My OwN ReSeArCh” and reached his own conclusions. The only peace in that man’s soul was that which he had with himself.

Bus Bob looked like he wanted to go back to any moment that didn’t have “Conspiracy Bro” anywhere near it.

“What do you think he does?” The Hat Man leaned over to ask me, studying Conspiracy Bro the way you study a weird error message you’ve never seen before.
“Nothing good,” I said. “Has that energy.” I replied.
“He looks like that one dude at the flea market who resells knives he got on TEMU...” The Hat Man postulated.

We chuckle and nod together in silent agreement.

My friend, who we will call “The Hat Man” and I have known each other for quite some time. We call him that because not unlike “The Hat Man” you see if you take too much Benadryl by mistake, he’s a little intimidating at first. Tech attracts all kinds, and it is what makes our community truly something special. He’s one of my few remaining links to my early days. Like me, he’s an old timer in the Enterprise IT space, and has had his fair share of adventures.

The Hat Man and I were doing what we always do in transit, half listening, half “somewhere else entirely” when Conspiracy Bro’s voice cut through everything like a car alarm.

“Carney’s a Davos guy, bud. FULL globalist. He’s not working for Canada, he’s working for THEM. The central bankers. All connected, I’m telling you right now!”

Bus Bob was nodding in the way people nod when they want someone to stop talking.

Hat Man slowly turned to look at me. Not his whole head. Just his eyes. The look that means “You seein’ this nonsense?”

He stared at Conspiracy Bro for an uncomfortably long time. Long enough that I thought he might join the conversation for the lolz.

I knew this guy. Not him specifically, but the shape of him. You hang around Upstate NY long enough and you will encounter this particular specimen at a bonfire eventually. The absolute confidence of a man who has done his own research. Their loudness fills the immediate area like they are the main character, and certain in the way that only comes from never having been seriously challenged by actual facts. The cognitive dissonance required to simultaneously believe the global elite are controlling everything and that he alone has figured it out is impressive. He wore it like a comfortable jacket.

He had found his way to Kelowna somehow, and I’m hoping that his stop is coming up soon.

“...and the AI thing, that’s THEM too, bud. That’s the control mechanism. That’s how they’re gonna...”

Hat Man’s head came up slightly at that.

“Your CUNY kids,” he said, pivoting instantly, the way he does, like a man running seventeen parallel processes. “They nervous about AI?”
“Differently nervous than him.” I replied.
“Good nervous or bad nervous.” He asked.
“Realistic nervous,” I said. “Like will there be a job for me nervous. Not...” I gestured in the general direction of Conspiracy Bro, who was still holding court.

Hat Man made a sound like a man who has bitten into something unexpectedly sour. Dug around in his jacket pocket for something, found nothing, seemed unsurprised.

“We survived the dot com crash,” he said, suddenly very quiet.

“We survived outsourcing. We survived the cloud eating the on-prem IT whole.” He looked out the window at the $1.79/liter Esso sign sliding past.

“You know who survived all of it? The people who understand the fundamentals...”
“FUNDAMENTALS, MAN!” He repeated, louder still, as if he, too, was about to launch into a conspiracy theory of his own.

“The people who knew the plumbing underneath everything, and who could actually think.” He pointed vaguely in Conspiracy Bro’s direction without looking at him. “THAT guy is scared of a control mechanism. Your kids should be learning to BUILD the control mechanism.”

Conspiracy Bro laughed loudly at something. Still sure of himself, holding court at the front of the bus. The Hat Man watched him for a second, then turned back to the window.

“You tell your kids that,” he said - at a reasonable volume. “Fundamentals. Everything else is just noise with better marketing.”

The Hat Man is unhinged a lot of the time, and for some reason, banned from the Earl’s here in town - but he is, at the end of the day, terrifyingly smart. He put words to a thought that I have had since this whole Generative AI craze started. The fundamental building blocks of technology are where career longevity lives. And, in my inaugural dispatch, I want to address what that means in a rapidly changing technical landscape.

The 3 Skills That Survived Every Tech Crash Since 2000

Every time the tornado of “aggressive market changes” touches down and devastates the neighborhood, there are always a few stubborn invariants that fight that hateful vortex, and win. Some houses on your street are missing roofs, walls - someone’s whole tool shed is where your kitchen once was - but there’s always “grandma’s old couch” that you and your partner argued over how ugly and heavy it was when you first moved in that house - still stubbornly sitting exactly where you and your buddy who helped you lug that thing in there put it, even more stubborn than your grandma ever was. While you’re taking a break from cleanup, that is where you sit and have a drink. The “Grandma’s old couches” of our industry are what I am here to give to you today.

Linux: The Load-Bearing Wall

Linux is running everything you touch and half the things you don't. The public cloud workloads powering everything from Netflix queues to clinical trial data? Yeah, 90% of all that is some variant of Linux (source). The little embedded computer in your car telling the engine what to do? Linux. The router, the IoT sensor, the web server, the AI data pipeline currently feasting upon your personal data at a geometric rate? That runs on Linux too. Windows still has the enterprise desktop and Active Directory locked up, but the infrastructure underneath all of it, the actual load-bearing walls of the internet, is a Linux box or VM somebody configured, somewhere, probably with a cup of coffee going cold next to them. Understanding how that works, even at a surface level, even just the command line and the basics, puts you in a conversation that most entry-level candidates cannot have. That is a sharp edge that will keep your resume from getting the spike.

Here's where I'd tell someone I actually like to start:

• The Linux Foundation - Introduction to Linux (LFS101) - Right now, this is free (as of 3/13/2026). It’s a full course on the Linux operating system that you can take online - and it has labs where you can break shit. The Linux Foundation also has certifications you can cram for and stick on your resume.

• Coursera’s Linux+ Prep Course - The Linux+ is CompTIA’s Linux certification. This course is geared toward people who have messed around with Linux on their own and are looking to get certification exam ready. Free as of 3/13/2026. Most hiring managers in tech focused roles are familiar with the Linux+, so there is wide industry recognition of this cert. Even if you don’t sit for the exam, this prep course will set you up to sit for other general Linux certifications.

• Boot.dev’s Learn Linux - I have heard some good things about boot.dev - a few of the YouTubers I like to rot my brain to occasionally call this out in their videos. It is “gamified.” I sat and messed around with it a bit. It feels like playing some sort of RPG. I signed up for free, but there are some paid tiers that unlock more features on boot.dev. “Sort of free” as of 3/13/2026.

Networking: The Part Developers Pretend Isn't Their Problem

A junior developer I knew at an agency I worked at spent a full, billable, “could have shipped something” day troubleshooting connectivity issues on an AWS VPC configuration. The app wouldn’t talk to the on-prem resources through the VPN. People depending on this junior to ship code were getting antsy and frustrated because he was “critical path.” The problem was a subnet address scope overlap that anyone with a working understanding of how networks are addressed would have caught in about four minutes before writing a single line of configuration. A whole day gone. All because nobody told him it was his problem to understand. By the time I was made aware of it, it had grown to mythical proportions and was being reported to me as an “Our cloud is down!” problem.

Here’s the thing about modern development nobody puts in the job posting: you are expected to know the network impact of your code now. The wall between “developer” and “infrastructure person” has been quietly demolished by cloud architecture, and many junior candidates are standing where that wall once was, confused about why the rubble is their problem, and quietly questioning their career choices.

The flip side of that story is also true. The junior candidates who walk into a room knowing this stuff have a specific kind of confidence that’s immediately visible. Troubleshooting or scoping an application that needs to scale doesn’t rattle them because they aren’t secretly terrified of the networking piece. That calm is a tell. Hiring managers notice it even when they can’t name it.

The AI can’t save you here either. Not because it doesn’t know what a subnet mask is - it knows exactly what a subnet mask is, and it will tell you with complete confidence, in a very helpful tone, while being subtly wrong about your specific situation in a way you won’t catch because you don’t have the mental model to know what question to ask it in the first place, or to recognize the wrong answer when it arrives neatly formatted in a markdown code block. It gets worse at scale. When you’ve got a team of developers all independently asking AI to help with network architecture decisions, you’ve distributed the problem without solving it. The AI doesn’t have the full picture of your infrastructure. Neither does the developer in the next sprint asking a slightly different question. What you end up with is a lawless gathering of locally reasonable answers that nobody ever looked at as a whole, until something breaks prod, and the ticket comes in as “Our cloud is down.”

Which is why I'd tell you to study for and pass the Network+. It won’t make you a network engineer - that’s not the point - but it will give you enough of a foundation to not burn a day on a subnet mask. If you did decide to become a network engineer, though, it would give you a critical foundation upon which you can pivot.

Here’s where I’d start:

• TechWorld with Nana: Every Networking Concept Explained In 20 Minutes - This woman’s YouTube channel helped me level up my Python skills 5 or so years ago. She breaks down critical networking concepts in a “here’s what you need to know right now” way, making her a great first stop.

• Tech Gee’s 5 Hour Full Course - It covers everything you need to know to pass the exam. CompTIA’s exam objectives can be found here if you want to see what you’re getting into first.

• Coursera’s Networking Fundamentals: This course includes hands-on labs and is free as of 3/14/2026. This covers some networking and security topics. Akamai is the core sponsor of this course. They are a large cloud provider like Cloudflare, and someone you will see a lot in the CDN space (“Content Delivery Networks”).

Python: The Ceiling You Don't See Until You Hit It

Nobody gets hired into a junior role and immediately handed a Python project. That’s not how it goes. Here’s how it actually shakes out: You get hired, you’re doing well, you’re picking things up, and then - somewhere around month three or six - you run into a wall that nobody warned you about. An Ansible playbook needs to be modified, or a Terraform module that needs to be modified to deploy a new resource to prod. Or, my personal favorite - a data pipeline is throwing an error that the senior engineer is too busy to look at. Today, though, you are more likely to see an AI-generated script that’s supposed to automate something critical and needs somebody to verify it actually does what the ticket says it does before it gets deployed to prod.

Every one of those moments assumes you can, at the very least, read Python and understand what it’s doing well enough to reason about it, modify it, and eventually own it. The people who can’t do that quietly stop growing and eventually get routed around. The interesting problems go to someone else, and that gets noticed by your managers and peers.

Python is the one place where AI is genuinely useful. Claude, Copilot, ChatGPT, Gemini, they all write decent Python, and you should absolutely use them. But “use them” is not the same as “trust them blindly.” If you can’t read the code they generate, you are not a developer using a tool - you are a liability with a keyboard. At the end of the day, you are the one who has to explain what that script does when it breaks something at 2am, and “the AI wrote it” can be a career meteor when it comes time to do the postmortem.

Python is the native tongue of everything that's eating the world right now - AI tooling, data pipelines, automation, cloud infrastructure scripting - you absolutely need to understand it if you want to have a future in tech since it serves as the basis of everything important today. In the same way that Excel spreadsheets are the language of the global financial system, so too is Python the glue language that holds a lot of modern IT stacks together.

Here’s where I’d start:

• Python in 100 Seconds by Fireship.io - I absolutely love this dude’s channel. He explains Python in 100 seconds with examples, and memes.

• TechWorld with Nana’s Python Course - This is the course that helped me write better Python. The video is 5 years old, and since then, there have been many updates to the language, but the fundamentals still hold up. Nana doesn't understand the huge, profound impact she had on my career, so I wanted to call this one out specifically. I wouldn’t be where I am today without her.

• Visually Explained Learn Python for Beginners - This whole playlist on YouTube is more current and uses Google Colab for running the code you do along with her.

The Stop Comes Up Fast

Conspiracy Bro was still loudly filling the air with his nonsense when we got off the bus. I don’t know if Bus Bob ever made it home or just quietly disintegrated somewhere around the third stop out of sheer exhaustion. The Hat Man said nothing else of consequence - just pulled his collar up against the cold the way he does and disappeared in the direction he came from, the way he always does, without a formal goodbye.

But that line stayed with me the whole walk home. Fundamentals. Everything else is just noise with better marketing.

The noise right now is very loud and very well funded. There are a lot of people with a lot of money invested in making you feel like the ground is shifting too fast to stand on, that the only move is to chase whatever thing just got announced last Tuesday. That feeling is a product. It is being deliberately sold to you. It is, in its own way, a control mechanism.

The floor is still the floor. In this case, Linux is still that floor. The network is still the network, and Python is still the glue holding the whole stack together. None of that changed last Tuesday, and none of it will change next Tuesday, either.

You are not behind. You are not too late. You are standing at the beginning of a career in one of the only industries on earth that will pay you to keep learning for the rest of your life, but only if you learn the right things first.

Start with the floor.

Untill next time - Be well, friends!