attention head AI jobs Alerts Saved
AI jobs › Model evaluation AI jobs

Model evaluation AI jobs

Model evaluation is how a team knows where a model stands. Training a new model or changing a product's prompt is only worth it if someone can show the result is better, and public benchmarks stop telling models apart once most of them score well. So much of the work is designing new tests: hard questions with checkable answers, realistic tasks for agents, and measures of behaviour that has no single right answer.

Research scientists design evaluations of capabilities and risks and study what scores really mean, including whether test questions leaked into training data. Research engineers and ML engineers build the harnesses that run evaluations throughout training and the tools that track the results. Applied AI engineers build test sets from real use and grade outputs with rubrics, people or other models. Expect statistics, care with data, and the judgement to tell when a number is misleading.

We count roles whose postings mention evals, model evaluation, or evaluation frameworks, harnesses, suites or pipelines.

1,239 open AI roles at 714 employers, from their own job boards. Showing the newest 100.

Follow Model evaluation97 new this week

New Model evaluation roles by email, the morning after we find them, in one email with your other alerts. Narrow it, or keep every role.

How often
We email a link to confirm it.
Anthropic · 37Amazon · 20NewRocket · 17ServiceNow · 14NVIDIA · 13LangChain · 10Libra Technology GmbH · 10Mercor · 10Microsoft · 10NCS · 10Articul8 · 9EverAI · 9Waymo · 9Anyone AI · 7Cursor (Anysphere) · 7

Get new Model evaluation roles every Monday

A job alert: new Model evaluation roles from employers' own job boards, every Monday. Unsubscribe in one click.

Related