Work Out What Kind of AI Role This Really Is
When to use it
Turn a JD with 'AI' in every other line into a clear answer on which talent pool you're actually hiring from.
The prompt
This job description mentions AI, machine learning or LLMs. Tell me what kind of role it actually is before I start sourcing. {{Paste the JD}} TITLE ON THE REQ: {{e.g. AI Engineer, ML Engineer, Applied Scientist, GenAI Developer}} WHAT THE HIRING MANAGER SAID IN THEIR OWN WORDS: {{if you have it}} Classify it against these roles, which draw on different talent pools and pay differently: - Research scientist: creates new models or methods; usually a PhD and publications. - ML engineer: trains, tunes and deploys models, and owns the data and training pipeline. - AI engineer / AI application engineer: builds products on top of existing models (APIs, RAG, agents, evals) and rarely trains anything. - Data scientist: analysis, experiments and classic predictive models for business decisions. - ML platform / MLOps engineer: the infrastructure other ML people run on. - A normal software role with some AI features bolted on. Give me: 1. Your classification, with the three or four lines of the JD that decided it quoted back. 2. Where the JD contradicts itself, such as a research-scientist wish list on an application-engineer budget, or "build agents" alongside "PhD required". 3. The two or three requirements that genuinely matter for this role, and the ones that are buzzwords that crept in. 4. The titles people doing this work actually use on LinkedIn, so I search the right pool. 5. The three questions to take back to the hiring manager to confirm or overturn your classification. If the JD is too vague to classify, say so rather than picking the most likely answer.
Tip
Most mis-hires in AI start here: a team that needs someone to build products on top of existing models writes a research-scientist JD, then wonders why every candidate is too expensive or too academic. Settle the classification with the hiring manager before the first outreach goes out.
How to use it
Paste the prompt into ChatGPT, Claude or whichever assistant you use, then replace every {{bracketed}} part with your own detail. The more specific and messier your input, the better the output — a model given raw notes has more to work with than one given a tidy summary you wrote first.
More Technical Translation prompts
Open the full tool for all 87 prompts, the glossary, and the JD decoder.