AI hiring is booming. AI Engineer roles grew 143% year over year. Companies are competing hard for talent.
But here’s the problem. Most of them are hiring wrong.
They post for “Senior ML Engineer” or “AI Researcher.” Then you look at the actual work. It’s connecting an OpenAI API to their existing product. That’s it.
Most companies don’t need someone who can fine-tune models or train neural networks from scratch. They need someone who can integrate AI into their workflows. Someone who understands how to present AI features to users in a way that actually helps them.
A product engineer. Not a researcher.
I see this pattern constantly with startups. They burn through runway hiring expensive AI specialists when all they really need is to connect their data as RAG to the chatbot they just built.
There are three tiers of AI work. Most companies are at Tier 1 and hiring for Tier 3.
Tier 1: API Integration. Connecting OpenAI, Claude, or Gemini APIs to existing products. This is where most startups live. You need a full-stack engineer with AI integration experience. Not a PhD.
Tier 2: RAG and custom pipelines. Vector databases, retrieval systems, data ingestion. This needs data engineers and NLP specialists. Still not a researcher.
Tier 3: Fine-tuning and multi-agent systems. Custom model training. Multi-agent orchestration. This is where you actually need ML engineers and AI researchers. Very few companies are here.
Companies can’t find what they need because they don’t know what they need.
If you’re a founder or CTO about to hire your first AI person, ask one question first.
Are we building models or building on models?
The answer changes everything about who you should hire.