Hey dev community 👋
I’m a web developer, and lately, I’ve been feeling a bit stuck on where my career is heading.
Where I started vs. Where I am now
- Before: My daily workflow was hands-on: writing code, reviewing PRs thoroughly, and testing features. The absolute most challenging—and interesting—part of my job was actually writing the code itself.
- Now: I go into the office, skim/review PRs, test implemented logic, and spend most of my day chatting with Claude/AI tools to execute my tasks.
While AI has made me vastly more productive, it has replaced the core part of the job I used to find most fulfilling. I feel like I'm moving away from deep technical craftsmanship, and I’m honestly not sure what direction I’m heading toward or what my role will look like in the next 3–5 years.
My questions for the community:
- Skill shift: What should I focus on studying or building to ensure long-term career growth in an AI-heavy workflow?
- Role evolution: For developers in similar situations, how are you shifting your focus? (e.g., system architecture, product engineering, devops, soft skills)?
- Fulfillment: How do you stay technically sharp when AI is handling the heavy lifting for daily tickets?
I’d love to hear your thoughts, experiences, or any roadmaps you’d recommend!
Top comments (2)
Great questions!
I think you should focus on studying how to build on your own apps rather than relying on AI to build them for you because it tends to write junior-level code.
To stay technically sharp—even if you choose to use AI to write code for you—don't focus on the 'how to do this,' but the 'why this works.' Understanding the logic behind the code is the absolute best way to stay technically strong, because coding is 10% syntax and 90% logic.
Honestly. Take the spare-time to turn yourself into a niche expert. AI makes mistakes, so the 'fullstack developer' role is on it's way out, because anyone can use AI. Your new role is to be an expert in a specialized field, so you find the errors AI makes, that nobody else can. Eg. CyberSecurity.
I find that while people write AI slop, they are using AI wrong. You can blow a million dollars on a team's API bills, to get a functional app, or you can blow 100k on building frameworks and tooling that lets the AI build that app for a fraction of that. Take V.A.L.I.D. I wrote it to replace CSLA for managing business objects in Blazor (C#). Then I expanded it to generate practically everything with Roslyn that isnt unique. Now it saves me 82% of all code I need to write, but more importantly it handles the tough stuff, like Unit Tests, MCP-wiring, Business Objects, Rules, Data Access Layers, UI wiring, etc. All I do is define a DTO with ValidObjects tags and restraints and write the UI markup. It generates everything else. So I wrote a skills file for it, so the AI knows how to take an existing module and convert it to V.A.L.I.D. and result is dramatic cost-savings, with robust non-slop guaranteed code and it performs 6x better than traditional Blazor, even with multi-parameter rule-checking. Use AI to build the tools to make the job more efficient.
You shift your focus from being a grunt to being a manager. Your job isnt 'write code' anymore, it's 'orchestrate your team of LLMs to boost productivity and lower costs'. People who will see themselves out the door are the ones that just focus on productivity and think having AI write 3000 LOC in 10 minutes is good enough. It's the old give a man a fish, vs teach them to fish saying in practice. AI can give you the 3000 LOC and you do the same pattern daily, by day 10, you'll have spent 30k LOC worth of tokens and every day you add, you bloat your consumption more. Or you can spend 30k LOC worth of tokens building frameworks that turn that 3k LOC task into 300 LOC. The real value isnt the token savings, it's the fact that you can always trust the framework to do it's job perfectly, LLMs dont even trust their own code and neither should you. So focus your efforts on infrastructure that makes the job faster, easier and cheaper. That's likely to get you a promotion, instead of fired for unsustainable API bills.