A few years ago, the "Prompt Engineer"—someone who knew the magic words to whisper to the model—was the hottest role in tech. As models become smarter and more instruction-following, this specific skill is depreciating.
The Challenge: The gap between Data Scientists and App Developers.
Traditional Data Scientists build models. Traditional Software Engineers build apps. But who builds the glue? Who decides when to call a tool? Who architects the evaluation pipeline?
Recommendation: Hire for "Agentic Orchestration."
The defining role of this era is the AI Systems Engineer. This is a full-stack developer who understands:
- Orchestration: Chaining multiple calls and managing state.
- Evals: Building automated scorecards to measure regression.
- Tool Use: Defining clean JSON interfaces for the model to interact with the world.
Hire for architectural thinking, not just prompt crafting.