I’m writing this from my desk on the day after Christmas, looking back at a year that felt like three years compressed into one. I started 2022 at a large consumer platform and a real-time messaging company, building infrastructure and helping teams ship. I’m ending it watching half the industry lay people off while a chatbot rewrites the assumptions we all had about what software development looks like.
What a year.
The Work
My main focus this year was a large consumer platform and a real-time messaging company, and the contrast was instructive.
At a large consumer platform, I worked on infrastructure that had to be reliable at scale – food delivery doesn’t tolerate downtime during lunch rush. The engineering culture was strong, the problems were interesting, and the systems were complex enough that good decisions compounded visibly. Working on their platform gave me a front-row seat to what happens when infrastructure is treated as a product rather than a cost center. It works. Teams ship faster, incidents are contained better, and new engineers become productive sooner.
At a real-time messaging company, the challenge was different – real-time messaging infrastructure at massive scale, where latency and reliability are the product. The work forced me to think carefully about state management, edge cases in distributed systems, and what “good enough” looks like when your customers are building their own products on top of your platform. It also reinforced something I keep learning: the boring engineering decisions – how you structure your Terraform, how you manage state, how you test infrastructure changes – matter more than the clever ones.
Both experiences reminded me why I like variety in my work. You see more patterns, more organizational shapes, more ways things can go right and wrong.
The Layoff Wave
I watched the layoffs arrive from a strange angle – close enough to feel the impact, removed enough to see the patterns across multiple organizations.
The speed of the pivot was jarring. Companies that were hiring aggressively in January were cutting in November. The same leadership that pushed “grow at all costs” pivoted to “efficiency is everything” without acknowledging the whiplash.
The organizations that handled it well were honest about the business reasons, treated departing people with dignity, and immediately rescoped their roadmaps to match the smaller team. The ones that handled it poorly did multiple waves, pretended the plan hadn’t changed, and burned out the remaining team trying to deliver the same roadmap with fewer people.
I’ve been on the founding side of this. When Voymigo’s runway got tight, I had to make similar calls. It’s never clean. But there’s a difference between hard and careless. Some of what I saw this year was careless.
ChatGPT Changes the Conversation
Then in late November, ChatGPT landed. I wrote about my first impressions already, but looking back at the full year, the trajectory is clear: AI went from a curiosity to a daily tool in about twelve months.
I started the year barely thinking about AI in my day-to-day work. Copilot was interesting but not essential. I ended the year using ChatGPT multiple times a day for drafting, debugging, and thinking through problems. The speed of that adoption – in my own workflow, not just the industry – caught me off guard.
The biggest shift isn’t productivity. It’s the realization that code generation is becoming commoditized. The cost of producing a first draft is dropping toward zero. What remains expensive – and will stay expensive – is knowing whether the draft is correct, appropriate, and maintainable. Judgment. Context. Taste.
For someone who has spent a career building expertise in systems design and infrastructure, this is actually reassuring. The skills that matter most are the ones that AI can’t fake yet: understanding tradeoffs, designing for failure modes, knowing when “good enough” is the right answer.
What Held Up
Looking back across the year, a few principles proved durable:
Discipline over heroics. I keep coming back to this. The teams that survived 2022 in good shape weren’t the most talented or the most well-funded. They were the most disciplined. Clear priorities, sustainable pace, honest communication.
Boring infrastructure wins. Every Terraform module I wrote this year was boring on purpose. No clever abstractions, no over-engineered wrappers. Simple modules with clear inputs, versioned state, automated tests. The boring stuff is what you can trust at 2am.
Relationships compound. The work I got this year came from people I had worked with before who trusted me to deliver. No amount of marketing replaces that.
Looking at 2023
I expect the layoff wave to continue into Q1 at least. The companies that cut early and honestly will recover first. The ones that are still pretending everything is fine will have a harder Q1.
AI will keep accelerating. By mid-2023, I expect code generation to be a standard part of most engineering workflows, not an experiment. The teams that will benefit most are the ones investing in verification – testing, code review, and quality infrastructure – not just generation speed.
For my own work, I plan to keep building, keep contributing to Go, and keep writing. This year reinforced something I believe strongly: the fundamentals don’t change. Languages change, tools change, economic cycles change. But clean architecture, honest communication, and sustainable engineering practices outlast all of it.
2022 was hard. It was also clarifying. Sometimes you need the music to stop to figure out who was actually dancing and who was just standing near the speakers.