We are building an industry that cannot replace itself
7 min read
A mentee of mine recently told me he had applied to over 400 software engineering roles. He has a computer science degree, a solid portfolio and genuine enthusiasm for the craft. He got 10 interviews. None of them gave him constructive feedback on why he was not progressing.
Four hundred applications. A 2.5% interview rate. And not a single company told him what he could do better.
When I graduated in 2008, I sent about a dozen applications and still nearly changed career into IT support. But I was competing with other graduates. The playing field, while difficult, was at least level. That is no longer the case.
The landscape has shifted
A computer science graduate entering the job market today is not competing with other graduates. They are competing with everyone.
Bootcamp graduates who spent three to six months building and shipping real applications. Self-taught engineers with strong GitHub portfolios and live projects. AI-native developers who have been using large language models as part of their workflow for years. And increasingly, experienced engineers who have been made redundant and are now applying for the same entry-level roles.
This is not speculation. Entry-level developer hiring has collapsed by approximately 67% since 2022. Software development role postings on Indeed fell 71% between February 2022 and August 2025. Computer science graduates now face a 6.1% unemployment rate, among the highest for any degree subject. The number of graduates entering the market grows every year while the roles available to them are shrinking.
The junior developer pipeline is not just competitive. It is structurally broken.
AI is reshaping the market from both ends
The pressure on graduates is coming from two directions simultaneously.
From above, AI is enabling senior engineers to do more with less. A senior developer with agentic tooling can now deliver output that previously required junior support. Companies are noticing. Over half of surveyed organisations have reduced or stopped hiring junior developers specifically because of AI productivity gains. Why hire a graduate when your existing team just got faster?
From the side, AI-driven redundancies across the tech industry displaced over 100,000 workers in 2025 alone. These are not junior engineers. They are experienced professionals with years of delivery behind them, now entering the same job market as fresh graduates. When a hiring manager has a choice between a graduate with coursework and a mid-level engineer with production experience, the outcome is predictable.
I wrote previously about how AI is changing the shape of engineering work. The agentic SDLC demands stronger pipelines, better specifications and engineers who understand systems deeply. That shift creates real opportunity for engineers who can think beyond code. But it also raises the bar for entry at precisely the moment when getting through the door has never been harder.
The feedback vacuum
This is the part that frustrates me most.
If you interview a candidate and decide not to progress them, tell them why. It is that straightforward. A graduate who has spent months applying to hundreds of roles and getting silence in return has no way to improve. They do not know if their CV is the problem, their technical skills, their communication or something else entirely.
My mentee had a nice graduate CV. Clear, technically broad, showing varied projects across different languages. But it was not competitive against the market he was actually in. The profile section was generic. There was no mention of AI-native development, no frameworks, no deployment experience, no testing tools. The project descriptions said what was built but not how it worked or why it mattered.
I could see that because I have twenty years of context. He could not see it because nobody told him. Ten interviews and not one piece of actionable feedback.
Companies will argue that legal risk prevents them from giving feedback. That is largely a convenient excuse. You do not need to write a performance review. A single sentence would change someone’s trajectory. “Your technical skills are strong but we needed to see more evidence of building end-to-end systems.” That costs nothing and it gives a candidate something to work with.
If your hiring process treats candidates as disposable, do not be surprised when the talent pipeline dries up. It is already happening.
What graduates and entry-level candidates can do
The market is difficult but not hopeless. Graduates who position themselves well can still stand out. The key is understanding what hiring managers are actually looking for and it is no longer just a degree and some coursework.
Show that you can build and ship. Coursework demonstrates learning. Deployed applications demonstrate capability. Get two or three solid projects hosted with live URLs. Talk about the process, the iteration, the tools used. Show pull requests, pipelines and a workflow that mirrors how professional teams actually deliver software. End-to-end understanding of build, test, deploy and monitor is what separates a candidate from a student.
Demonstrate AI fluency, not just AI usage. “Used AI” is not a differentiator in 2026. Everyone is using AI. What matters is showing understanding. Which models did you use and why? What prompting strategies did you apply? How did you use AI beyond code generation, for planning, testing strategy, code review and security analysis? A computer science graduate has the theoretical foundation to use AI more effectively than most. That is an advantage worth showing.
Show engineering thinking, not just coding ability. Hiring managers want signals that you understand how software is built professionally. Mention your testing strategy. Reference performance considerations. Show security awareness. Discuss trade-offs you made and why. These demonstrate a depth of understanding that a portfolio of code alone does not.
Be specific about collaboration. Vague references to team projects do not tell a hiring manager much. What methodology did you use? What was your Git workflow? How was work planned and how were decisions made? Specificity signals experience. Generality signals inexperience.
Improve your language. Less “developed X” and more “what did I build, how did it work and why does it matter?” The difference between “Developed an AI chatbot using Python” and “Built a Python-based chatbot to manage restaurant bookings using LLMs via the OpenAI API, designing prompt flows to maintain context across multi-turn conversations” is the difference between a list of tasks and a demonstration of capability.
The long-term problem nobody is talking about
Here is the uncomfortable reality that the industry is ignoring.
People like me will retire. Every senior engineer, every staff engineer, every principal and every engineering manager will eventually leave the industry. That is not a risk. It is a certainty. The question is who replaces us.
If companies stop hiring graduates now because AI makes senior engineers more productive, they are solving a short-term efficiency problem by creating a long-term talent crisis. Junior engineers become mid-level engineers. Mid-level engineers become senior engineers. That pipeline takes years. You cannot skip it and you cannot start it later without consequences.
The industry is optimising for the current quarter. Cut junior headcount, boost productivity metrics with AI tooling, report efficiency gains. But a team of senior engineers augmented by AI is not a sustainable model if nobody is coming through behind them. In five to ten years, organisations that stopped investing in early-career talent will face a senior engineer shortage that no amount of AI tooling can solve. You cannot prompt your way to architectural judgement. You cannot generate institutional knowledge. You cannot automate the mentorship that turns a graduate into someone who can lead a team through a production incident at two in the morning.
Companies need to think beyond the next sprint. Hiring graduates is not charity. It is infrastructure. It is how you build the engineering capability that will carry your organisation forward when the people who built it are no longer there.
This is a choice
The graduate pipeline is broken and it is getting worse. AI is compressing the market from both ends. Experienced engineers are flooding into roles that used to be reserved for fresh talent. Companies are not hiring junior engineers and when they do interview them, they are not telling them how to improve.
None of this is inevitable. Companies can choose to invest in early-career engineers. They can provide meaningful feedback to candidates. They can recognise that the short-term efficiency of a smaller, senior-heavy team comes at the cost of long-term organisational resilience.
Graduates can adapt too. The bar is higher than it has ever been but the tools available to meet it are better than they have ever been. A graduate who can demonstrate AI fluency, engineering thinking and the ability to ship working software is still a compelling hire. The opportunity exists. It just demands more than it used to.
But this is not a problem that candidates can solve alone. The industry built this situation. The industry has to be part of fixing it.
We cannot keep pulling up the ladder and then wonder why nobody is climbing.