When you’re hiring in AI, it’s easy to be drawn to the obvious things on a CV.
A recognisable company name. An impressive job title. Experience with the right models, tools and technologies. A degree from a leading university. A list of projects that seem to tick every box.
But some of the strongest candidates aren’t always the ones who immediately stand out on paper.
AI is a particularly difficult market to judge through a CV alone because job titles and responsibilities can vary significantly between companies. An AI Engineer at one business could be building and deploying models into production, while someone with exactly the same title elsewhere could be working on a completely different part of the stack.
The same applies to the projects candidates list.
There’s a big difference between being part of a team that built something and being the person who identified the problem, made key technical decisions, dealt with things going wrong and ultimately took responsibility for delivering it.
A CV doesn’t always show that distinction.
And as AI teams become increasingly specialised, understanding what someone has actually done becomes even more important. Two candidates could both have LLM experience, for example, while having completely different levels of involvement, technical depth and responsibility.
That’s why the conversation behind the CV matters.
Grace, who works within Odiin’s AI team, sees this regularly when speaking with candidates.
“I’ve spoken to candidates who looked fairly average on paper and then completely changed my mind within 15 minutes. A CV can tell you where someone has worked, but it doesn’t always show you what they really owned, how they approached problems or the impact they had. That’s usually the detail that comes out in conversation, and sometimes that’s where the strongest candidates really stand out.”
That word — owned — is particularly important.
Ownership can be difficult to communicate in a few bullet points. Did someone simply execute an existing plan, or did they help decide what should be built in the first place? Were they given a solution to implement, or did they have to work out the solution themselves? When something didn’t work, were they responsible for figuring out why?
Those differences tell you far more about how someone operates than a list of technologies ever could.
Conversation also gives candidates the opportunity to explain the context behind their experience.
Perhaps they worked at a company nobody immediately recognises, but were given far more responsibility than they would have received at a larger organisation. Maybe their job title undersells what they actually do. They might not have worked with the exact technology listed in a job description, but they’ve solved remarkably similar problems elsewhere.
None of those things necessarily jump off the page.
It works the other way too.
An impressive CV can open the door, but it doesn’t automatically tell you how someone thinks. It doesn’t tell you how curious they are, how they communicate a complicated idea, how they react when challenged or whether they genuinely understand the technology they’ve worked with.
Those qualities are becoming increasingly important in AI.
The technology is moving quickly enough that today’s highly sought-after technical skill could look very different in a few years. Knowing a particular tool or model matters, but so does someone’s ability to learn, question, adapt and solve problems they haven’t encountered before.
That creates an interesting challenge for hiring teams.
The more specific a search becomes, the easier it is to turn a job description into a checklist: this title, this many years of experience, these companies, these technologies.
Checklists make filtering easier. But they can also filter out people who don’t look like the obvious candidate.
And sometimes those are exactly the people worth speaking to.
This doesn’t mean experience, technical ability or a strong CV suddenly don’t matter. They do. A CV is still an incredibly useful starting point for understanding someone’s background.
The problem comes when it becomes the entire assessment.
Especially in a market as fast-moving as AI, perhaps the better question isn’t simply whether someone’s CV matches the job description.
It’s whether the person behind it could.
