The Market Doesn’t Want Juniors Anymore

Over a decade ago, when I was starting in IT, it wasn’t easy to land that first job. The golden era for juniors ended in the second half of the 2010s. Despite that, businesses still needed people eager to learn, and some hired young talent. Some left after a few years, while others stayed much longer.

With the emergence of AI, seniors have become much more capable. A smart senior who delegates writing to AI and keeps their time free for thinking and decision-making can achieve much more. And AI is flexible. You can directly tell it when something isn’t working. No need to spend time providing „positive feedback”, lookin for the „right words”. If somethings is bad - you name it. AI doesn’t care, it doesn’t get upset or offended. There’s no friction between the overworked senior and the rest of the team as there used to be.

Every Senior Was Once a Junior

Nobody joined the market as a senior. An IT job requires so much knowledge that it's impossible to learn it all in any university. The best way to learn is to start building.

Building alone early on is tough. Many questions, few meaningful answers. If someone wasn’t comfortable with trial and error, learning was nearly impossible. The solution was finding a good deal: low salary with great support and mentorship. Mentors were like compasses for juniors - navigating and helping deliver smaller, non-urgent features. But these were real deliverables.

Some developers never realized they were being mentored and guided, even if “senior” was in their title. These were knowledgeable and reliable workers, that needed someone to tell them what to do. Sometimes that was a golden cage.

Switching Googling to Prompting

Back then in 2010’s, the crucial skill was googling - using search engines for getting the information fast. With the right query, you could find discussions, answers, examples, docs, and code snippets. Anything that gave you clues about how to solve a problem.

The modern world has changed. Now, you need to be good at prompting AI. It means AI does the search for you and prepares the explanation. In most cases, it’s more efficient than searching manually. But the main challenge remains.

Using Solutions vs. Learning from Them

I’ve observed this for years: people build feature after feature, switch teams, and make the same mistakes. Write the same inefficient code. Count months and years to claim their seniority. All that while making mistakes that stem from the most popular answers on dev forums or examples copied and pasted across internal sources.

I was fortunate enough to always ask "why?" Others were content with having a working solution. It was a shortcut to marking a problem solved and moving on.

Now, with AI assistance in code writing, it’s even easier. No need to copy-paste. "Vibe coding," right? Everything is fine - until it’s not. The endless unconscious decisions can lead to a catastrophe.

Who “owns” these vibe-coded codebases? The maintainer badge isn’t about having the power to merge directly to the main branch. It’s about making long-term decisions - thinking, analyzing, and expressing doubts. It’s about acknowledging the imperfection of oneself and fostering growth in others. Endless learning and self-navigating that was always the super skill of great seniors. Being curious about a better solution, advocating for building software the better way.

Ban the Vibe Coders

Anything fun is fine, as long as it’s not harmful. Generating utils or apps with AI can be great. It offers a lot of flexibility with a fraction of the time investment. But we need to develop good practices to prevent the harm of a vibe-driven approach.

Data leakage, wasted effort, unmanageable codebases - these problems are after a dozen wrong decisions. Is reading all the generated code the right (or only) solution? I’m not sure. It’s fun to have a working feature or app, but it’s not fun spending hours fixing tiny bugs caused by a lack of understanding of the codebase.

How to Make the Future Brighter?

We must reconsider who junior developers are and what they’re capable of. Coding and learning APIs are no longer problems. Making good decisions is. And it’s not fair to expect juniors to make those decisions. That’s the responsibility of seniors and architects. They aren’t titled "senior" or „architect” because they know 50 ways to validate a string. Their value lies in choosing the right approach from those 50.

What can juniors provide these days? What should they learn to fit into the modern software development landscape? Mastering AI prompting is a meta skill, but making smart decisions and organizing code the right way? That’s too early for a junior. Should juniors be treated as long-term investments, spending 3-6 months learning? Capitalism won’t like it, as others may complete 3-6 projects in the same time.

Since there are no seniors without juniors - we must find an answers to these questions. Coding tasks don’t show much anymore with assisted interviews and generated codebases. Could this be a chance for the IT space to become more about humans? A chance to focus on soft skills? How do we ensure there are enough hard skills when necessary?

What do you think? What should be the requirements and learning path for the modern junior software engineer?