35 questions to sniff out a misleading AI engineering role

Interview Preparation •

Frontier AI models are upending generative roles, and software engineering is no exception.

Ninety-one per cent of respondents to a 2026 GitLab poll said they currently used two or more AI coding tools. Such tools are making coding faster, according to 99 per cent of respondents to Atlassian’s 2025 Developer Experience Report.

As AI tools alter the day-to-day experience for engineers, many job hunters enter an entirely new environment when stepping into a role. As teams navigate a new way of operating with AI, growing pains are being felt.

Some candidates seeking AI engineering jobs on Bravou reveal contradictions between what job descriptions claim and the reality of their role. That’s causing some to step away from hard-won jobs early in their term.

How does a job-seeking AI engineer avoid a role that isn’t what the job description claims? By interviewing armed with a specific list of questions.

This article recommends the questions to ask, categorised by the specific topics you might want to explore. We also spotlight the employer responses to listen for so you can avoid an AI engineering role that isn’t what it seems.  

Quick tip

You wont be able to ask all 35 of the questions below. Choose which to prioritise based on:

  • Clues hidden in the job description. This article can help you identify those clues.
  • Who's interviewing you. The hiring manager can talk about strategy. Another engineer is best placed to explain your day-to-day tasks.
  • The dealbreakers that matter most to you. Decide in advance which three factors would make you decline the role and look out for them.
  • The type of employer. A startup calls for questions around budget and flexibility. For a larger organisation, talk through senior buy-in and data access.

1.       The reality of the role

  • What would I release in my first 90 days, and who is it for?
  • Roughly what split of my time goes to building new systems, maintaining existing ones, cleaning data and stakeholder work?
  • What did the last person in this role (or the person on the closest team) spend most of their usual week doing?
  • Is this role building models or integrating and orchestrating models someone else built?
  • What is the most recent thing this team put into production, and how long did it take?
  • Is this a new seat or a backfill? If a backfill, why did the last person leave?


Listen for: Vague answers such as “you’ll be doing a bit of everything”. This often points to a maintenance-heavy role with few opportunities to build systems and tools of any strategic value.

2.       Data

  • Who owns data quality and pipelines and is there a dedicated data engineering function?
  • Is the data I’d need already accessible, labelled, and documented, or would I be sourcing and cleaning it?
  • How long does it take to get access to a new dataset?
  • Are there data residency, privacy or sovereignty constraints that limit what I can use or where I can run workloads?
  • How is data quality monitored once a system is live?


Listen for: Named owners and existing pipelines. If they don’t exist, it could mean you’ll also be the data team.

3.       Budget and resources

·      What’s the compute budget for this team? Who approves extra budget?

·      Do I have access to GPUs or cloud credits on demand, or do I share the capacity with others?

·      What are the monthly inference or API costs today and is there a cap on them?

·      Is there budget for tooling, evaluation infrastructure and third-party data?

·      Is the team’s budget ring-fenced or reallocated each quarter?


Listen for: Specific figures or ranges. If the interviewer can’t provide this, it’s possible the company’s AI claims are more marketing spin than a true strategic priority.

4.       Team and structure

·      Who would I report to and what is their technical background?

·      Who else is on the team and what are their roles (ML, data, platform, product)?

·      Who would I work with day-to-day outside the team?

·      Is the team growing and what roles are planned next?

·      Where does the AI team sit in the organisation, and who is the team’s most senior champion?

·      Are there other AI engineers I can learn from, or would I be the only one?


Listen for: Clear reporting lines and a team that supports the company’s stated AI ambitions. If key roles are missing, the business’ claims might not align with what the team is currently structured to deliver.

5.       Priorities and strategy

·      What are the top two or three priorities for this team this year?

·      How are those priorities set and how often do they change?

·      Who sets the priorities for this team?

·      How does this work connect to revenue, cost savings or a strategic goal?

·      Is the AI roadmap driven by a business problem or by a company-wide mandate to use AI broadly?


Listen for: A short, defined list or priorities backed by a logical rationale. Too many top-level priorities suggests a lack of focused direction and a lot of experimentation that leads to nothing meaningful.  

6.       Production and engineering practice

·      What does the deployment process look like and how long is it typically from idea to production?

·      How are cost, latency and model drift monitored?

·      Is there an on-call or support expectation, and how common is it?

·      How much technical debt is there in existing AI systems and is time budgeted to address it?

·      How do you handle vendor and model changes (for example a provider changing pricing or deprecating a model)?


Listen for: Evidence of mature development practices. Use your experience to decide what sounds like a building role versus a team that struggles to get past the demo stage with projects.

7.       Leadership and organisational buy-in

·      How does leadership view AI: as a core capability or an experiment?

·      What example can you give of a time the team asked for investment in infrastructure or data and received it?

·      What happens when an AI project needs support from a function that has other priorities?

·      Has any AI project been cancelled, and why?


Listen for: Specific examples with supporting evidence, such as quantifiable results data. It’s easy to claim leadership support, but proof comes from real evidence.

The final word

 It’s wise to view the interview process as a fact-finding mission to understand the reality of the role, alongside your main priority of impressing the employer. This is particularly true given the extent AI is changing engineering roles.  Armed with some of the questions above, you can get the clearest picture possible before you’re stuck in a role you didn’t sign up for.

Searching for your next AI engineering role in the UAE, Saudi, Bahrain, Kuwait, Qatar, or Bahrain? Click here to explore them all on Bravou.