How to Hire AI Engineers when EVERYONE is cheating with AI — Beth Glenfield, DevDay

The Broken State of Technical Hiring 00:00

  • The speaker, Beth Glenfield, introduces herself and addresses the challenges AI poses to traditional technical hiring.
  • Questions are posed about using AI in recruitment, competing with tech giants for candidates, and the effectiveness of coding puzzles like LeetCode.
  • Glenfield claims technical recruitment is now "very broken", with small companies losing out in the talent war.

How AI Has Changed the Landscape 01:23

  • Over the past 18 months, there has been a rise in AI-enabled cheating services for technical interviews, such as Clu, which raised $5.3M and nears $1M ARR.
  • Success rates on LeetCode-style coding challenges are extremely high (up to 93%) among candidates interviewing at companies like Google and Meta.
  • About one in three interviews now involve candidates using AI assistants.
  • The current interview process often assesses who has the best AI assistant, not who has the best engineering skills.
  • Industry leaders note the shift: Sam Altman advocates learning AI tools, and Salesforce reported a 30% productivity boost after replacing some engineers with AI (though Glenfield questions the data).
  • Job competition is no longer only about pay but also about company brand, prestige, and stability, especially amidst frequent layoffs.
  • Top candidates are more likely to choose major tech divisions over startups due to perceived security and reputation.

Skills Needed for Modern AI Engineering 02:53

  • Ideal candidates for AI development are creative, collaborative, and adept at working with AI, not just skilled at coding challenges.
  • These candidates build AI tools, use AI libraries, contribute to open source, and understand business impacts rather than focusing solely on coding puzzles.
  • LeetCode and similar approaches no longer measure the most relevant skills for actual job performance.

Rethinking Technical Interviews 03:24

  • Glenfield suggests observing candidates’ collaboration with AI on real-world business scenarios, rather than on abstract algorithm puzzles.
  • New assessment methods could include evaluating delegation, ambiguity management, and adaptability to changing requirements.
  • Demonstrating a company's engineering culture and simulating real work environments during interviews can help assess better fit.
  • At DevDay, the approach emphasizes workplace simulations where candidates work alongside diverse AI agents (e.g., perfectionists, pragmatists, security experts, juniors needing mentorship).
  • These simulations place candidates in authentic settings, requiring day-to-day tradeoffs relevant to the company’s business domain.
  • Evaluation focuses on collaboration with AI, communication (e.g., pull requests, tickets), decision-making, mentoring, and adaptability.

Impacts on Companies of Different Sizes 05:17

  • Large companies like Google or Meta can "brute force" hiring, screening many candidates and offering high salaries to a select few.
  • Smaller companies face high costs for bad hires, with losses ranging from $20K to $60K, and cannot afford to hire engineers who can’t ship AI products.

The Future of Engineering Jobs 05:53

  • AI may soon handle most mid-level engineering work, with reports suggesting entry-level roles are being eliminated.
  • While the nature of engineering jobs is changing, there will remain demand for roles emphasizing creativity, collaboration, business judgment, and working with AI.
  • DevDay collaborates with design partners interested in evolving their hiring practices and invites further discussion on this approach.