I was recently laid off. My first reaction was dread about having to enter the job market again.
Part of that dread was financial. I had a savings plan for things I want in the future, like owning a home. I was also hoping to travel. Now I have to think more carefully about those plans because I don't know how long finding another job will take.
More than anything, the layoff interrupted my sense of momentum. After working professionally for more than two years, I felt like I was starting to build a stable career and plan for what came next. Now I am trying to figure out where I fit again.
That question feels more complicated than it did when I first started. I have enough experience to understand what goes into shipping real software, but I am still early in my career. At the same time, AI is changing what software development looks like and which skills feel valuable.
Two years is an awkward amount of experience
Over the past two years, I have worked on web and mobile applications, often in healthcare. One project I am especially proud of was a web dashboard for caregivers. I built most of the frontend myself, but my role went beyond that.
I contributed to the backend and deployment, interacted with clients, and had opportunities to make product decisions and suggest improvements. I liked having that kind of ownership. The product had a meaningful use in healthcare, so the decisions I made felt connected to something real.
That experience helped me realize that I am good at solving problems. It also showed me what I want from my next role. I don't want to complete tickets without understanding why they matter. I want to have input on the product and feel like the work I am doing will actually help someone.
Two years is an awkward amount of experience in the current market. I am not brand new, but many roles ask for more experience than I have had time to build. There are also so many people applying that getting a resume in front of a real person can feel like a completely separate challenge from proving that I can do the job.
On top of that, the skills companies need from developers are changing.
I spent years learning how to code. Now someone with no coding experience can use AI to spin up a convincing MVP. That doesn't mean the result is reliable, maintainable, or able to scale. But it does make me wonder how the skills I spent so much time developing fit into software going forward.
I think understanding and validating what was produced matters more now.
When the developer starts directing the work
AI has made me a much faster developer. I use agents for implementation, initial reviews, and iterations.
I usually start by laying out the architecture I want and adding the constraints I can think of. I ask the agent to push back when something is unclear or when there is something I might not have considered. Then I delegate the implementation and initial review. After that, I review and test the result myself. I keep going through that loop until I am happy with it.
In that workflow, I sometimes feel like I am directing the work more than doing the actual coding.
There are a lot of things I like about that. I can test ideas and build things much faster than I could before. At the same time, it can be hard to know where my own skills fit when an agent is producing so much of the code.
I don't think using agents automatically makes someone an architect. Architecture requires an understanding of how a system works, where its faults are, and how it can scale. That understanding usually comes from experience. It is difficult to evaluate an implementation if you have never struggled through one yourself.
People often say that developers just need to adapt to AI. I agree, but that doesn't really answer the question for me. What should I actually spend my time learning? The tools are changing so quickly that it is hard to know which skills will still matter a year from now.
I think this creates a difficult problem for early-career developers. A lot of our judgment comes from trying something, getting it wrong, and figuring out why. If an agent handles most of the implementation, it can also remove some of the trial and error that helps us learn.
I don't think the answer is to avoid AI or write everything manually. I think developers still have to experiment and try different things. AI lets me test more approaches in less time, but speed only becomes experience if I stay involved in the loop.
I try to make sure I understand exactly what the agent is doing and how its solution could be improved. I inspect the result, test it, and question decisions I don't understand. If I can't explain how the important parts work, then getting a successful result isn't enough.
For me, that is where judgment comes from. You try something, see what works, find the problems, and try again. AI changes how fast that process can happen, but I still have to learn from it.
I think the differentiator going forward will be whether you fully understand the system that was built. Can you identify its faults? Do you understand how it will behave when the data grows or the requirements change? Can you recognize when the agent solved the wrong problem elegantly?
AI can review code and suggest architectures, but it can't take responsibility for what gets shipped. It also can't replace communication with clients and teammates. I think understanding the product, making good decisions, and taking ownership of the result are becoming more important parts of being a developer.
The job search is a human problem too
One of the most frustrating parts of the job market is how difficult it is to get a resume seen by a real person. When so many people are applying, it is tempting to send as many applications as possible and hope one gets through.
I am starting to think that building genuine connections is more useful than spamming applications. My first real software job came from a connection I had made when I applied to the same company while I was still in school.
That earlier experience did not lead to a job right away, but it created a relationship. Years later, that relationship turned into an opportunity. Someone had a better understanding of who I was than they could have gotten from a resume alone.
That has changed how I think about networking. I don't want to collect connections just so I can ask people for referrals. I want to share what I am building, talk to people in the industry, and give them a chance to understand how I think and what I care about.
That is part of why I have been working on my own projects, improving my website, becoming more active on LinkedIn, and writing this article. I can't control which companies are hiring or whether they see my application. I can keep building, make my work more visible, and start more conversations.
I am still looking for where I fit
I don't have a clear answer for where software engineering is going or exactly where I fit within it.
I still worry that AI could replace more of the work I currently consider valuable, including parts of problem-solving. I am still trying to decide which skills matter most and how someone early in their career can prove that they have them. I am also dealing with a job search I did not expect while reconsidering plans that had started to feel secure.
At the same time, I know I have something to offer.
I have taken meaningful ownership of a product. I have worked across frontend, backend, deployment, client conversations, and product decisions. I know how to use AI to make myself faster while still making sure I understand what it produces.
I also know what I want from my next role. I want to be part of a team where I can have a real impact on the product and the decisions behind it. I want the work to feel meaningful. I don't want to build another button that nobody will use.
I am still uncertain about where I fit, but I am learning how to move through that uncertainty. For now, that means continuing to experiment, continuing to build, and continuing to understand the systems behind whatever AI helps me create.