AI Product Engineer
Builds user-facing products on top of existing model APIs — the engineering around the model call, not the model itself.
Also posted as: AI Engineer · GenAI Engineer · LLM Engineer · Applied AI Engineer
Definition
What this role actually is
An AI Product Engineer builds products people actually use on top of existing model APIs. They don't train models — they turn them into something usable: the retrieval, the prompts, the fallbacks, the evals, the streaming interface, the cost controls.
The job is product engineering where the hardest dependency happens to be non-deterministic. Most of the difficulty is not getting a good output once. It's getting an acceptable output every time, at a price that works, and knowing when it didn't.
The line in the sand
What it is NOT
Not an ML Engineer.
ML Engineers train and fine-tune models. AI Product Engineers consume them. Different skill, different pay band, and the two are conflated constantly — often deliberately.
Not a Prompt Engineer.
Prompting is one skill inside this role, not a role of its own. A portfolio consisting entirely of prompts is not evidence of this role.
Not a full-stack developer who added an API call.
The distinguishing work is everything around the model call — evaluation, failure handling, cost and latency management. One integration is not the job.
Output
What they actually ship
- Production features with a model in the critical path — chat, generation, extraction, classification
- Eval suites that catch quality regressions before users report them
- Streaming interfaces, token and cost budgets, caching, retry and fallback logic
- Structured output pipelines that survive a model version change
Stack
Tool stack
Depth
The proof standard
Tier 1
Core
Shipped a product with a model in the critical path that real people use. Can show the eval suite, name a specific production failure mode they fixed, and state their cost per request.
Tier 2
Working
Built and deployed a live, reachable feature using model APIs. Handles errors and bad output. May not have formal evals.
Tier 3
Light
Built working prototypes or demos against model APIs. Not yet in production with real users.
Depth is what the builder claims. The badge is awarded separately, from the evidence — read the standard every badge is judged against.
Evidence
What a strong portfolio piece looks like
A live product with a public URL where the AI feature is the core value — plus a written breakdown of one thing that broke in production and how it was fixed. The breakdown matters more than the product. It's the difference between having shipped once and knowing the job.
Market
Market snapshot
The largest and most title-inflated category in AI hiring. Identical work is posted as AI Engineer, GenAI Engineer, LLM Engineer, Applied AI Engineer and Forward Deployed Engineer. 2026 analysis puts the US median for API-based product builders at roughly $159K–$245K, with title choice alone observed to swing offers by $80K for the same work.
Last updated: 2026-08-30