Hiring usually starts with a job description.
But some of the hardest startup hiring problems have very little to do with finding people who match one. The role may be defined too narrowly. The founder may know the outcome they need without knowing the exact profile that delivers it. Or a candidate may tick every technical box and still be wrong for the reality of an early stage company.
That’s where recruiting stops being search and starts being about the judgment.
We spoke to Pooja Kumari, a startup tech recruiter at Uplers, about how she approaches these situations. One principle keeps showing up across the roles she’s worked on:
Before finding the right candidate, make sure you’re solving the right hiring problem.
A JD is where the conversation starts, not where it ends
When Pooja gets a new requirement, she doesn’t start searching. She starts asking why.
Why does this hire need to happen now? What will this person actually own? What would make them successful six months in? Which skills are truly non negotiable, and which ones simply made their way into the JD out of habit?
That distinction can change a search entirely.
One startup came to Uplers looking for a frontend engineer, with Vue listed as a core requirement. The problem: it was shrinking the talent pool significantly. The easy response would have been to keep hunting for more Vue engineers. Pooja questioned the premise instead.
Did the company really need someone who already knew Vue? Or did it need a strong frontend engineer with solid JavaScript and TypeScript fundamentals who could pick up Vue quickly? That reframing opened the search to strong React engineers with the right underlying capabilities and those profiles started progressing.
The lesson wasn’t that founders shouldn’t have specific requirements. It was that specificity only helps when it predicts success in the role. Sometimes a search is hard because the talent is genuinely rare. Sometimes it’s hard because the talent has been defined too narrowly. A good recruiter knows the difference.
But knowing when to narrow the search matters just as much
Pooja saw the opposite problem while working with our client MathsOnline, which was hiring senior developers against an unusually nuanced bar. Technical ability mattered, but so did employment stability, genuine enthusiasm for coding, and the ability to operate well in a remote, cross cultural team. The challenge wasn’t finding more senior developers it was understanding exactly which senior developers MathsOnline would actually consider.
So Pooja looked backwards. She studied past candidate decisions: who progressed, who got rejected, and what reasons kept showing up in the feedback. What was the client consistently evaluating that never appeared in the JD?
Patterns emerged. She used AI to organize those patterns into a screening framework, then took it back to the client for validation. She’d captured almost their entire evaluation logic they made one adjustment. Then came the real test.
3 profiles submitted. 3 interviews. 2 offers.
The improvement didn’t come from access to better candidates. It came from understanding “right fit” far more precisely which points to a second principle in how Pooja recruits:
Every rejection should make the next shortlist better.
If five strong looking candidates keep failing at the same stage, sending five more won’t necessarily fix anything. The pattern itself is information.
Then comes the part a JD can never capture
Even with the requirement perfectly calibrated, one question remains: will this person actually thrive in a startup?
Pooja has seen technically exceptional candidates who were simply wrong for the environment they were entering. So her assessment doesn’t stop at skills and experience she wants to understand how someone thinks about ownership, ambiguity, change, and motivation.
If a candidate says they want the ownership of an early stage startup, does their track record support that? If they say they’re comfortable with ambiguity, where have they actually operated in it? If they say they want a small company, but every other role they’re seriously considering is at a large enterprise, what are they really looking for?
Pooja reduces it to three questions:
- Can they do it? that’s capability.
- Do they genuinely want it? that’s motivation.
- Will they thrive here? that’s environment fit.
Capability is usually the easiest to assess. The other two are where startup hires tend to succeed or fail.
Where AI fits into all of this
AI has changed how Pooja works through these questions. She uses it to understand unfamiliar technologies, analyze profiles, spot patterns in rejection feedback, build screening frameworks, and prepare sharper technical questions. The MathsOnline search is a good example of what that unlocks.
But there’s an important line she holds: AI gives her another perspective, it doesn’t get the final vote. If an AI recommendation says reject and her own assessment says otherwise, she investigates the disagreement rather than defaulting to the model’s output.
AI is becoming very good at answering who appears to match this requirement? Startup hiring still needs someone to ask:
- Is this the right requirement in the first place?
- What isn’t this CV telling me?
- Why do good candidates keep getting rejected at the same stage?
- Does this person actually want the environment we’re offering?
Those are judgment calls. And as finding candidates gets easier, Pooja believes that judgment becomes more valuable, not less.
What founders can take from Pooja’s playbook
Three ideas worth carrying into your next hire:
1.Treat your JD as a hypothesis, not a specification. The market and the candidates you meet should sharpen your understanding of the role, not just fill a template.
2. Treat rejection as data. If good candidates repeatedly fail, understand the pattern before asking for another batch.
3. Separate capability from startup fit. “Can they do the job?” is only one part of “Should we hire them?.
None of this requires a founder to become a recruiter. But it does change what you should expect from one.
A great startup recruiter shouldn’t just understand the role you’ve written down. They should understand your company well enough to challenge the requirement when it’s wrong, sharpen it when it’s unclear, and recognize the right person when a CV alone can’t tell you.
Because the goal was never to put more profiles in front of a founder. It’s to make every profile that reaches them worth their time.
