OpenAI's newest team is proof that there's a key shift happening in the AI talent wars (www.businessinsider.com)

OpenAI has created a new Applied Evals team, led by Shyamal Anadkat, to help business customers build rigorous, domain-specific evaluations (evals) for complex, real-world tasks — think refund processing, code migration, voice AI and multi-step reasoning. The group will work with OpenAI’s sales and developer teams to prioritize projects where models underperform, and is hiring engineers with “real, deep, lived expertise” (roles paying roughly $255K–$325K plus equity). The team will start with generalists and then add specialists (e.g., software engineers for code tasks, and potentially humanities experts for writing-focused use cases). The move highlights a broader shift in the AI talent wars: companies are no longer just recruiting model-builders but also domain experts and eval engineers who can define “what good looks like” for specific applications. Technically, this signals growing emphasis on nuanced, context-aware evals and domain-specific benchmarks (beyond binary “thumbs up/down” tests) to operationalize LLMs safely and reliably in production. It also underscores tighter integration between research and product teams to close gaps between capability and real-world usefulness — a trend likely to raise demand (and compensation) for practitioners who can translate model behavior into dependable business outcomes.