I ran CantonCoders.org, a non-profit that helped over 80 people transition into tech jobs. We never charged a thing. Back then, the path was clear: get a certification or degree, practice interviewing, build a portfolio. It worked. You got a job.
That path from 2023 doesn’t exist anymore. And honestly, neither does the path that brought you to where you are today.
You’ve worked five, ten, maybe fifteen years to get here. You’re good at what you do. You’ve earned your position. I’m writing this because I care about your career thriving, not just surviving, and what I’m seeing across the industry needs to be said directly.
The Reality Right Now
Organizations have rolled out AI coding agents and are watching what drives dramatic business improvements and who stays flat. Many organizations are going all-in on AI. The ones that aren’t won’t help you remain relevant in 2028.
If you’re at a pure software company (SaaS, platforms, developer tools), you have 12 to 18 months before patterns become clear and start driving promotion decisions.
If software supports a physical product or service (manufacturing, healthcare, logistics), you have three to five years. But the gap between people building these capabilities now and people waiting is widening every quarter.
The Market Truth Nobody’s Saying
Right now, companies can’t hire someone with three years of applied AI development experience because those people don’t exist yet.
In 2028, they will exist.
When that company is hiring for senior, staff, or principal roles, are you the person with the track record, or the person being passed over for someone who has it?
This isn’t about job security. This is about being marketable for better opportunities in 2028. About having a resume that opens doors instead of raising questions about why you didn’t adapt.
Why This Is Hard for Us
We built tacit knowledge over years through osmosis. When we explain things to colleagues, we rely on shared context. We point at code and say “like this.” We’ve rarely had to externalize everything explicitly.
AI agents have zero shared context. When we can’t externalize our knowledge, we can’t work with them. They weren’t in the room in 2015 when we picked that database.
We also built navigation skills—”I know where things are and what patterns work”—sometimes without building mental models of why things work. Navigation made us productive. Agents need understanding to reason about what to build.
None of this is a character flaw. It’s a debt that’s come due.
What Your 2028 Resume Needs
In 2028, when you’re interviewing for that senior, staff, or principal role, they’ll ask: “Tell me about your experience with AI agents. How did you adapt? What business improvements did you drive?”
Your competition will talk about reducing cost of goods sold, increasing flow, improving time-to-market. They’ll have concrete business metrics. They’ll talk about building AI systems, orchestrating agents, establishing governance in the SDLC.
Will you have those stories? Or will you say “My company rolled out tools but I mostly worked the way I always did”?
The Six Capabilities You Need to Build
1. Build Externalization
Explain your thinking to people with zero context. Write documentation that transfers understanding, not just describes what exists. Do pair programming where you make your reasoning visible. Present technical decisions to your team.
The discomfort you feel doing this is the skill building. It’s like going to the gym—uncomfortable while you’re building the muscle, but that’s how you know it’s working.
2. Build Mental Models
Don’t just know that things work—understand why. Read code without tasks. Draw diagrams for yourself. Build the understanding you might have skipped on the way up.
Ask yourself constantly: do I understand why this works, or just that it works? Can I explain the architectural decisions, or do I just know they exist?
3. Practice Differently with AI Agents
Treat agents like mentoring someone capable but context-free. Write out what you’re trying to accomplish before giving instructions. If you struggle to articulate it clearly, that’s feedback about your understanding gaps.
When output surprises you, figure out what you failed to specify. That’s your externalization gap showing.
4. Learn to Build AI Systems, Not Just Use AI Tools
This is critical. The SDLC is fundamentally changing. You need to understand how to build systems that incorporate AI agents as components. How to orchestrate multiple agents working together. How to handle the unique challenges of testing, monitoring, and debugging AI-enabled systems.
The developer who can only use an AI coding assistant is different from the developer who can build systems that use AI agents as part of their architecture.
5. Understand Agent Orchestration
Systems in 2028 won’t just have one AI doing one thing. They’ll have multiple agents with different capabilities working together. You need to understand how to design systems where agents collaborate, how to handle coordination between them, how to manage state and context across multiple agents.
This is a new skill that almost nobody has yet. Build it now.
6. Learn Agent Governance in the SDLC
How do you test code that an agent helped write? How do you review it? What’s the right governance model for agent-generated code in your deployment pipeline? How do you ensure quality and security when agents are part of your development process?
Understanding this will separate engineers who can work safely and effectively with AI from those who create risk.
The Timeline Reality
You have twelve months to start building the track record that matters for 2028.
Engineers who started a year ago already have stories to tell. They can talk about specific systems they built using AI agents. They can show business outcomes: “I reduced the cost of our data pipeline by 40% by building an AI system that handles routine transformations.”
Engineers who start today will have solid experience by late 2026. They’ll have built multiple AI systems. They’ll understand agent orchestration from real projects. They’ll have governance frameworks they actually use.
Engineers who wait another year will be competing in 2028 against candidates with three years of demonstrated success building AI systems, orchestrating agents, and governing AI in their SDLC—candidates who can point to concrete business outcomes they drove.
What Success Looks Like
By 2028, you should be able to walk into any interview and tell stories like:
“I built a system that orchestrates three different AI agents to handle our code review process. One agent checks for security issues, another validates architectural patterns, and a third handles test coverage analysis. We reduced review time by 60% while actually improving catch rates.”
“I established the governance framework my team uses for AI-generated code. We have clear policies on what agents can touch, how we review their output, and how we test AI-enabled systems. Our velocity increased 45% while maintaining our quality bar.”
“I built mental models of our entire platform architecture by documenting it for an AI agent. Turns out I didn’t really understand some parts as well as I thought. Now I can explain any component to someone with zero context, and my ability to work with agents improved dramatically.”
These are the stories that get you the senior, staff, or principal roles. These are the capabilities that make you valuable regardless of what specific AI tools exist in 2028.
This Isn’t About Age
I’ve seen 50-year-old engineers thrive with these tools and 30-year-old engineers struggle. Fresh graduates and veterans both fail and excel.
The correlation isn’t age. It’s whether you built certain capabilities during your career: externalization and deep system understanding.
If you spent your career explaining technical decisions to non-technical people, you probably have externalization. If you built deep mental models and stayed curious about “why,” you probably have understanding.
If you were the person who “just knew” things and navigated intuitively without explaining, you have gaps to fill.
Age doesn’t determine your category. Your career path does. And these are learnable skills at any stage.
Your Career in 2028 Starts Today
The market in 2028 will have developers with proven track records building AI systems, orchestrating agents, and governing AI in the SDLC. Developers who can walk into interviews with compelling stories, real metrics, demonstrated capability.
Those developers will get the opportunities—the senior roles, the staff positions, the principal engineer jobs at the companies you actually want to work for.
You can be one of those people. But you need to start building that track record today, while companies still develop people internally, so you have it when the market has options and you’re competing against people who already spent three years building theirs.
You can do this. The capabilities are learnable. The path is clear enough if you start walking it.
Start today. Where you are three years from now—what opportunities you have, what companies want to hire you, what your trajectory looks like—depends on what you do in the next twelve months.
Engineering leader who still writes code every day. I work with executives across healthcare, finance, retail, and tech to navigate the shift to AI-native software development. After two decades building and leading engineering teams, I focus on the human side of AI transformation: how leaders adapt, how teams evolve, and how companies avoid the common pitfalls of AI adoption. All opinions expressed here are my own.