Red flags on an AI specialist's CV
7 minute read. Updated 2026-08-08.
The useful CV red flag is a specific tell paired with a resolving question, not a reason to reject outright. Watch for CVs with no failures, titles that don't match described tasks, long tool lists with no depth, and outcomes with no named decision attached. Gaps between contracts, short engagements and a non-AI degree are not red flags; they are normal features of fractional work.
A red flag is a question, not a verdict
Most hiring guidance treats a red flag as a reason to reject a CV outright. That's a mistake with technical hires, because the person reading the CV usually isn't technical enough to know whether what looks wrong actually is. The useful version of a red flag is a specific tell, paired with the one question that resolves it in a five-minute conversation.
The list below is written for a CTO or head of data who has to shortlist AI specialists without being one themselves. Each item names what to look for, why it matters, and what to ask. A separate section covers the things that look like problems and aren't, because treating those as disqualifying loses good candidates for the wrong reasons.
Every project is a success
A CV where every listed project ends in a positive outcome, shipped, adopted, improved, with no mention of anything cancelled, pulled, or wrong for a period, is worth a second look. Production AI work fails regularly: models drift, evaluation catches a problem after launch, a project gets cancelled when the data turns out not to support it. Someone who has done this work for years and has nothing to say about any of that has either not done much of it, or is choosing not to talk about it.
The resolving question: 'Tell me about a model or system you shipped that you later had to pull or fix. What went wrong, and how did you find out?' A genuine practitioner answers this quickly, with specifics: what broke, how it was caught, who caught it. A vague or defensive answer tells you more than the CV did.
The title doesn't match the work
A CV titled 'AI Governance Lead' that lists no register, no audit trail, no contract review, only meeting attendance and policy documents, is a mismatch worth checking, because nothing is published for that title in the UK advertised market and the role is being defined differently at every company that uses it. The same applies to 'AI Engineer' roles that turn out to be entirely prompt writing with no evaluation or monitoring behind them.
This is not automatically a problem. Titles are inconsistent across the market precisely because employers are inventing them as they go. But a title with no concrete activity behind it is worth checking rather than taking at face value. Ask: 'Walk me through what you actually built or reviewed in your last two roles, day to day.' Listen for specific systems and specific decisions, not job description language repeated back.
Breadth with no depth
A skills section listing a long, undifferentiated run of tools and techniques, large language models, retrieval-augmented generation, agents, fine-tuning, MLOps, vector databases, every current framework by name, without a single one of them appearing again in the project descriptions, usually signals a CV built to pass a keyword search rather than to describe what someone has done.
The resolving question is simple: pick one item from that list and ask them to go deep on it. 'You've listed fine-tuning. Tell me about the last model you fine-tuned, what data you used, and how you knew it had actually improved.' Someone who has done the work goes deep without hesitating. Someone padding a list starts speaking in general terms about the field.
A wall of certifications
A CV with 10 or more certification logos, cloud platforms, AI tools, prompt engineering badges, and comparatively thin project detail underneath each role, suggests time spent collecting credentials rather than delivering with them. Certifications are not worthless, but they measure completion of a course, not performance on a live system.
Ask which certification most changed how they approach a real piece of work, and why. A candidate who can point to a specific practice they adopted because of a specific course answers this in one sentence. A candidate who lists certifications as decoration struggles to connect any of them to something they actually did differently afterwards.
Outcomes with no decision attached
Phrases like 'improved efficiency' or 'delivered insights' without saying what decision the output fed into are a weaker version of the same problem. Every genuine piece of AI work changes what someone did next, approved a loan, flagged a transaction, routed a support ticket, and a CV that never names the decision is describing the field rather than the work.
Ask: 'What decision did that output feed, and what happened if it was wrong?' If they can't answer the second half, the project may have shipped without anyone checking what happens on failure, which is worth knowing about the candidate and about the project.
What looks like a red flag and isn't
Gaps between roles are not a red flag for a contractor or fractional specialist. A day rate includes no holiday, no sick pay and no paid time between engagements, so time between contracts is normal and expected, not evidence of someone struggling to find work.
A run of short contracts, three to six months each, is not job-hopping. It is how fractional and contract AI work is structured; clients bring in specific expertise for a defined piece of work and the engagement ends when that work does. Judge the work described in each contract, not the length of it.
No formal AI or machine learning degree is not disqualifying either. A meaningful share of strong practitioners in this field came from physics, software engineering, or statistics, and moved into AI work as it became a distinct discipline rather than starting in it. What matters is what they can show you they built, not the label on the degree.
What to do about it
- Treat a red flag as a question to ask, not a reason to reject a CV.
- Be wary of a CV with no failed or cancelled projects on it; ask what went wrong once.
- Check that the title matches described activity rather than taking it at face value.
- Pick one item from a long tools list and ask the candidate to go deep on it.
- Ask what decision an output fed into, and what happened when it was wrong.
- Do not penalise gaps between contracts, short engagements, or a non-AI degree.
Questions people also ask
Should I reject a CV that has no failures or setbacks listed?
Not automatically, but ask about it before shortlisting further. Ask the candidate to describe a project they shipped that later had to be pulled or fixed, and how the problem was caught. Genuine practitioners answer this readily with specifics. A defensive or vague answer, or a claim that nothing has ever gone wrong, tells you more than the CV's silence did.
Is a CV with lots of tools and frameworks listed a bad sign?
It depends whether any of them reappear in the project descriptions. A long list that never resurfaces once outside the skills section often signals a CV written to pass a keyword search. Pick one listed tool and ask the candidate to go deep: what they built with it, what data they used, and how they knew it worked. Depth on one item beats breadth across twenty.
Are gaps in employment history a problem for a contract AI specialist?
No. Contract and fractional work includes no paid time between engagements, so gaps are a normal feature of how this work is structured, not a sign the candidate struggled to find work. Judge the substance of what was delivered in each engagement rather than the tidiness of the timeline between them.
How do I check whether a candidate's job title reflects what they actually did?
Ask them to walk through what they built or reviewed in their last two roles, day to day, and listen for named systems and named decisions rather than job description language repeated back. Titles in AI are inconsistent across the market because employers are still inventing them, so the title alone tells you less than the answer to that question does.
Where the figures come from
Every rate and salary quoted in this article is a median or percentile of figures advertised in UK job postings over the six months to 8 August 2026. They are not rates paid, and the gap widens at the top of a range.
- IT Jobs Watch, UK contract rates, 6 months to 8 August 2026, read 2026-08-08.
- IT Jobs Watch, UK permanent salaries, 6 months to 8 August 2026, read 2026-08-08.