I'm based in the Balkans and have about four years of professional experience across frontend and backend development. My work has included rebuilding websites from Figma designs, creating APIs, integrating booking and availability systems, working with third-party services and payment providers such as Stripe, designing queries and indexes, improving performance with caching, and handling Docker, CI/CD, deployments, logs, configuration, and production incidents. I'm not a DevOps specialist, but I've worked around infrastructure and want to deepen those skills.
I'm comfortable working across the stack, but I know I have gaps in deeper backend concepts, system design, networking, infrastructure, and DevOps. I don't have a degree and I'm looking for a fully remote web development role, ideally with European working hours. I'm open to specializing in backend or infrastructure later, but I'm unsure whether to remain full-stack or go deeper in one direction.
I'd like to build a portfolio project without spending five months on a generic to-do app, CRUD application, or clone. I'm considering a backend-heavy project involving authentication, database design, payments, queues, caching, testing, CI/CD, and observability, but I'm unsure how much recruiters actually value a large portfolio project.
I'd appreciate practical advice on what companies look for after roughly four years of experience: how much a degree matters, which fundamentals I should know, whether I should focus on backend, DevOps, cloud, system design, algorithms, projects, or interview preparation, and whether AI tools should appear on my CV. I also want to understand how interviewers view AI-assisted or heavily AI-generated portfolio projects and whether "vibe coding" is considered a useful skill or a red flag. If you had six months to become substantially more employable, what would you prioritize?
3 Answers
Listing tools such as ChatGPT, Claude, or Copilot as standalone CV skills probably adds little because most developers are expected to use them. AI-assisted development is fine, including for a portfolio project, as long as you understand and can defend every important part of the result. Interviewers may question generated code more closely, so be ready to explain the architecture, security implications, edge cases, tests, and why you rejected or changed suggestions.
The valuable skill is not blindly prompting for an application; it’s giving an AI tool useful context, setting constraints, reviewing its output, testing it, and catching incorrect assumptions. That is worth demonstrating through concrete results, not by calling yourself a “vibe coder.” I also wouldn’t worry about AI replacing your production experience. Debugging a race condition in a payment flow or deciding how to recover from inconsistent external-system state still requires context and judgment.
Your strongest selling point is not a long list of technologies or a flashy project. It’s the real production experience you already have with payments, third-party integrations, booking flows, synchronization, performance problems, and incidents that cross frontend and backend boundaries. Present those as outcomes instead of responsibilities. For example, explain how you reduced a response time, handled failed webhooks, prevented inconsistent payment state, or fixed a production failure. A few concise case studies showing the problem, your decisions, tradeoffs, and measurable result will usually provide more signal than a huge GitHub project.
For a portfolio piece, build something related to a genuine operational problem. A small payment-webhook reconciliation service, for example, could demonstrate retries, idempotency, database consistency, alerting, observability, tests, and deployment without requiring an entire product. A focused booking service that handles concurrent requests and prevents double bookings would also show meaningful backend judgment. Keep the scope small, but make the difficult part excellent.
A project does not need to be large to be useful. Choose one difficult problem and document it thoroughly: the data model, failure modes, security decisions, testing strategy, deployment process, and tradeoffs. Interviewers can learn much more from why you chose a queue, index, cache, retry policy, or transaction boundary than from seeing ten unrelated features.
After four years, I’d prioritize practical fundamentals: SQL and database behavior, HTTP, authentication and authorization, concurrency, queues, caching, Linux, Docker, CI/CD, testing, observability, and basic cloud architecture. You don’t need to become an expert in every area, but you should be able to explain how systems fail and how you would investigate them. Practice system-design discussions using your own work, and prepare enough data structures and algorithms to pass the companies that test them.
That makes sense. I was mostly worried that openly mentioning heavy AI assistance would cause people to assume I didn’t understand the project. I’ll focus on documenting the decisions, tests, and failure cases rather than presenting AI usage as a separate skill.