[Feature Request] Add sample demonstrating how to make liteLLM calls from a workflow

[dandavison]

Add a sample demonstrating how to make LLM calls from a workflow via an activity.
This should use https://github.com/BerriAI/litellm.
(It may make sense to add something like this to SDK contrib/ eventually.)

[jssmith]

The AI Cookbook will contain this sample soon (and others). We could still put a version here, but it might not be necessary.

[Abhinav0905]

Hi @dandavison and @jssmith — before I start, could you confirm whether a samples-python version is still wanted now that the AI Cookbook may cover this?

I have hands-on experience combining Temporal, Python, and production LLM workflows. If the sample is still useful, I propose keeping it focused:

• Call LiteLLM only from Temporal Activities, not Workflow code.
• Demonstrate timeouts, retries, and workflow-safe inputs/results.
• Add deterministic tests using mocked provider calls, with no live API dependency.
• Include a README explaining why network I/O must remain outside Workflow code.

If this scope is aligned with what you want, could you assign the issue to me?

[brianstrauch]

Hi @Abhinav0905, feel free to open a PR. Tag me and I'll take a look when it's ready.