Last week, I spoke with a Talent Acquisition Manager who proudly showed me an AI-powered sourcing and resume review workflow he had built using Claude Code.
The concept’s impressive.
He had taken resumes from candidates who had previously been hired successfully, fed them into an AI workflow, and instructed the system to weigh incoming applicants against those past “successful placements.”
On paper, it sounded cool.
Then I asked a simple question:
“How does the scoring actually work?”
Silence.
I asked what the weighting looked like, how the matching logic functioned, whether the system was evaluating actual technical overlap versus keyword density, or whether it was just pattern matching resume phrasing.
He didn’t know.
He used AI to build something he couldn’t explain.
And honestly? That should scare people a little.
Because the conversation online right now has become:
“How do I beat the AI?”
But if the people implementing these systems don’t even understand the machinery themselves, candidates are trying to reverse engineer a black box that nobody operating it can properly explain either.
But this is not new, and that is the uncomfortable part.
Hiring was irrational long before AI entered the picture.
AI is not just creating dysfunction in recruiting. It is exposing dysfunction that already existed.
And despite what social media says, most companies are still not heavily relying on advanced AI systems for sourcing and application review. Some are experimenting. Some are automating pieces of workflow. Most are still operating through the same messy human systems that have existed for years.
The difference right now is that AI has lowered the barrier to intellectual curiosity so dramatically that ignorance no longer feels accidental or acceptable.
Ten years ago, maybe a recruiter genuinely struggled to understand technical nuance. Today you can open ChatGPT, Claude, or Perplexity and ask:
“What is the relationship between Angular and TypeScript?”
and get a clear answer in seconds.
That changes the equation.
And engineers deserve to understand the environment they are applying into.
Not so they can become cynical. Not so they can beat the ATS. But so they stop internalizing rejection as proof they are not qualified.
Because often, that is not what is happening at all.
I’m tired of candidates focusing on “how to beat AI” instead of focusing on themselves: their story, their value, and how to approach a job search more strategically. So I brought in another recruiting SME, Shawnee Malesich, and together we unpacked what we’ve both seen from inside large, big-box staffing organizations.
What started as a conversation about AI quickly turned into something broader. How hiring decisions get made behind the scenes, where qualified candidates get filtered out for reasons they never see, and why the gap between how hiring is described and how hiring works is still massive.
From that point on, this became a shared debrief between two people who have lived inside different parts of recruiting and have seen the same system break in all sorts of ways.
1. How Recruiters Actually Vet Resumes and Where Things Break
One of the first examples we both immediately landed on was a deceptively simple one.
A recruiter was hiring for a front end modernization role, Angular, TypeScript, enterprise rebuild.
A strong candidate came through with deep Angular experience and exactly the kind of modernization work the client needed.
The recruiter’s feedback came back:
“Good Angular experience, but not enough TypeScript. I’m passing on this resume. Go find me someone better.”
The issue?
Well. Angular development is primarily TypeScript-based.
What made this interesting is that this was not a rare edge case. Both of us had seen versions of this pattern repeatedly, where a candidate is screened out not because they lack capability, but because the person evaluating them does not fully understand the technical relationship between the skills they are screening for.
From the candidate’s perspective, this is where hiring becomes completely opaque.
There is no feedback loop that helps you improve.
Because the gap was not in your experience. It was in the evaluator’s understanding.
And that is where AI becomes relevant in a very different way than people assume.
It’s now something that makes basic technical ignorance easier to surface and harder to excuse.
2. The Part of Hiring Candidates Never See, Submission Decisions
This is where the conversation really opened up once we started comparing experiences.
Most engineers assume hiring works like this:
Apply → recruiter reviews → submit → interview
But in many staffing environments, there is a second decision point most candidates never see:
Recruiter → account manager → client submission
And that middle layer changes everything.
We compared examples where strong candidates were identified, validated, and ready to move forward, but still did not get submitted because of internal dynamics that had nothing to do with candidate quality.
One example was a QA role where a clearly stronger candidate was in play.
Technically aligned. Strong experience. Clean fit for the role.
But another recruiter in the same organization submitted a different candidate.
The deciding factor was not skill.
It was internal relationship dynamics with the account manager.
The weaker candidate got submitted.
The stronger one did not.
From the outside, none of that is visible.
The candidate only ever hears:
“You were submitted but not selected.”
Which is technically true, but completely incomplete.
Because the real decision happened upstream in a place the candidate never sees and can never influence.
We both landed on the same takeaway here:
Most rejection narratives are sanitized versions of decisions that were never actually about the candidate.
3. Why Engineering Resumes All Start to Look the Same
This was the other major overlap in our discussion, and one of the most consistent patterns across engineering hiring.
Resumes converge.
Same structure.
Same phrases.
Same technologies.
Same initiatives.
Same results driven engineer language.
And AI has only accelerated that convergence.
The biggest offender is what we both see constantly, the skills wall.
A dense list of tools, languages, frameworks, and platforms, disconnected from context.
From a recruiter’s perspective, it rarely clarifies anything.
If they do not understand your stack, it does not help them understand it.
If they do understand your stack, they already know how those tools relate.
What actually creates signal is context.
What was the product you built?
Who used it?
What broke if it failed?
What changed because you were there?
For example:
Built React components
vs.
Built authentication and checkout flows for a payments platform processing millions in annual transactions.
Same work.
Completely different signal.
And in hiring systems where attention is limited and interpretation is inconsistent, signal is everything.
Closing Thought: Stop Trying to Beat a System You Cannot See
The most important takeaway from our conversation was not about AI at all.
It was about how little of the hiring process is actually visible to candidates, and how much of it is shaped by human interpretation, internal incentives, and inconsistent understanding.
AI did not create that.
It just made it harder to ignore.
Which means the real challenge for engineers is not learning how to beat AI systems.
It is learning how to make themselves understandable in systems that were never fully consistent to begin with.
Clarity wins.
Context wins.
Signal wins.
Everything else is noise.
Shoutout to Shawnee @ CloudEmployee for jumping into this conversation and adding the kind of field level reality check that only comes from sitting inside staffing systems long enough to see how decisions actually get made.