Applied AI Engineer
Turn model capability into product behaviour that holds up outside a demo.
The gap between what a model can do and what a product can promise is where this role lives. Evaluation, retrieval, latency, cost and the failure cases nobody saw coming.
You will own the harness every product here is measured against, so a change that helps one product is a change that helps all of them.
What you would do
- Build evaluations that catch regressions before users do.
- Make the expensive thing cheap and the slow thing fast, without quietly making it worse.
- Keep a clear-eyed record of what each model is and is not reliable for.
- Say so when the honest answer is that a feature should not ship yet.
What we look for
- You have shipped model-backed features to real users and measured them.
- You are sceptical of benchmarks, including your own.
- You can read a paper on Monday and have something running by Thursday.
How to apply
hello@aiorg.ai
Subject line is filled in for you. Send a few lines on why this role and links to things you have made — shipped work, repositories, writing. No cover letter, and no form.