The Old Trap
For years, developers were told to pick a stack and go deep: "I'm a React developer", "I'm a Django person". Specialisation has real value, but it also created an identity problem. When the framework fell out of fashion, or the project needed something else, people felt stuck.
AI has changed the economics of that. The cost of working in an unfamiliar language or framework has dropped sharply.
What AI Changed
I have shipped production work in React, React Native, Next.js, NestJS, Express, and plenty of infrastructure tooling. These days I can pick up an unfamiliar framework and be productive in days, not months. AI handles much of the syntax, the idioms, and the "how do I do X here" lookups.
But notice what it does not handle: deciding what to build and how the pieces should fit together. That part is still yours.
What Actually Transfers
Frameworks are different surfaces over the same underlying ideas. If you understand these, you can move anywhere:
- ·Data structures and algorithms, enough to choose sensibly and spot a bad complexity.
- ·How the web and networks work: HTTP, caching, sessions, latency, DNS.
- ·Databases: modelling, indexes, transactions, and the trade-offs between relational and document stores.
- ·Concurrency and state: async, race conditions, and where state should live.
- ·Architecture patterns: layering, separation of concerns, boundaries, queues, event-driven design, monolith vs services.
- ·Security basics: authentication, authorisation, input handling, secrets.
- ·Testing and observability: how you know the system works and why it broke.
Why Fundamentals Matter More With AI
AI will give you an answer in any framework. It will not tell you when the answer is the wrong shape for your problem. Without fundamentals you cannot:
- ·Notice that the generated code ignores a failure mode.
- ·Tell whether a design will scale or collapse under load.
- ·Debug when the abstraction leaks, which it always eventually does.
- ·Choose between two reasonable-looking approaches.
The person who knows the fundamentals can direct the tool. The person who doesn't is being directed by it.
Architecture Is the Real Skill
Syntax is the cheapest part of software now. The expensive part is making good decisions: where the boundaries go, what is synchronous and what isn't, how data flows, what happens when a dependency fails. Those decisions look similar whether you are writing TypeScript, Go, or Python.
When I join a new project, I don't ask "do I know this framework?" first. I ask: *what are the components, what are the contracts between them, and where will it hurt?*
How I Approach a New Stack
1. Learn the mental model first. What problem does this framework solve, and how does it model it?
2. Map it to what I already know. Routing, state, data fetching, and deployment all have equivalents.
3. Build something small and real, with AI as a pair, but reading what it writes.
4. Learn the failure modes. Where does it break, and how do you debug it?
Closing Thought
Be curious, not loyal. Use the right tool for the problem, and let AI shrink the learning curve. But invest your real effort in the things that don't change: fundamentals and architecture. Those are what make you able to do anything.