LLM APIs
Also written as LLM API, OpenAI API, Anthropic API, Claude API, Gemini API, LLM Integration
The paid web services that let developers send text, images or files to a model such as GPT, Claude or Gemini and get a response back inside their own product. 'LLM integration' in a JD usually means building features on these APIs: prompts, tool calling, streaming responses, handling errors and rate limits, and controlling cost.
Think of it like
Renting a brilliant specialist by the minute over the phone instead of hiring and training your own.
Junior or senior?
The most common AI skill in JDs, so the call itself isn't differentiating.
Senior sounds like
Talks about what happens around it — evaluation, retries and fallbacks when the API is slow or down, caching, cost per user, and switching between providers.
Ask them
“What did you do when the model API was slow, down or returned something unusable?”
Sounds like real experience
Describes concrete handling: timeouts and retries, a fallback model, validating the output before using it, and tracking cost per request or per user.
Probe further if
Describes the happy path only — send a prompt, show the answer — with nothing about failures, cost or how they measured quality.