A candidate I was helping with applications this month said something that stayed with me. He is a 2023 graduate, a computer science degree from a college nobody has heard of, and he has been applying since before convocation. He has sent out, by his count, just under four hundred applications. He has received two callbacks. One was for a role that required three years of experience. The other was a sales job.
I did not have a comforting answer for him.
The headline numbers are striking: AI hiring is up twenty percent year on year, while overall white-collar hiring grew at two percent. Fresher hiring is essentially flat. This is not a recovery. It is a reallocation. Companies are not adding headcount broadly. They are moving money from general hiring into AI-specific roles, and they are doing it with the selectivity of someone who has been burned before.
Every week I talk to people who are in the middle of this. A data scientist at a large bank told me his team had a hiring freeze for nine months, and then suddenly got approval for three positions, all of them requiring prior experience deploying large language models in production. They could not fill them. The positions have been open for four months. Meanwhile, the general analyst roles on the same floor were eliminated in a restructuring.
This is what selectivity looks like. It is not that companies are not hiring. They are hiring for very specific things, and they are willing to wait for the right person. If you are a backend engineer with three years of experience and some exposure to ML pipelines, the market is better than it was last year. If you are a fresher with a degree and a few academic projects, the market is worse than it has been in a decade.
I spoke to a hiring manager at a mid-sized product company who put it plainly. He said they used to hire six freshers and two seniors for every team. This year they hired one senior and no freshers. The senior cost them more, but the senior could ship on day one. The freshers would have taken three months to become productive, and the company no longer believes it has three months to wait.
This is a structural shift, not a cyclical one. Companies that adopted AI tools over the last eighteen months found that their existing senior engineers could do more with less. The need for junior engineers to handle the grunt work diminished because the grunt work got automated. The pipeline that used to feed freshers into the industry, start with basic maintenance and debugging, learn on the job, graduate to real work in two years, has a narrowing mouth.
A recruiter at a GCC told me something that sounded brutal but honest. She said they used to hire freshers as an investment. Now they hire them only when a specific project demands it, and they have the budget. The rest of the time, the work goes to the seniors or gets done with AI tooling.
The candidates who are doing well in this market are not necessarily the ones with the fanciest degrees. They are the ones who can point to real work: a model they put into production, a pipeline they optimized, a system they built that actually runs. The degree is a checkbox. The work is what gets you hired.
I am not saying this to be grim. I am saying it because the fresher who sent four hundred applications deserves to know that the game has changed, and that sending more applications is not the answer. The answer is to find a way to do real work before someone pays you for it. Open source. A startup that will take a chance on you for low pay. A fellowship. A role that is adjacent to AI even if it is not the title you wanted.
The window for the old approach, good college, good grades, wait for the campus placements, has narrowed. It is not closed. But it is narrower than it was, and it will keep narrowing.
The candidate I mentioned at the start asked me what he should do. I told him to stop applying and start building. Pick something real, make it work, put it where someone can see it. It is not fair. But it is the move.

