Migrate Citation Generator

[jose-perales]

initial notes:

• currently in local script, migrate to notebook with appropriate triggers/access to datastores
• uses config to change questions in yaml for questions we want to ask/extract
• based on the type of docs, need to update the config

limitations:

• did not have enough data to experiment on optimal citation
• but, this is example code that can be swapped out with business logic for any use case
• we'll use LLM citation generation, but could also use search (RAG) or classification or different chunking strategies

Task List

• ensure migrated code works in databricks notebook
• migrate any Azure AI Services needed for Citation Generator

[raymond-nassar]

@joshuaphelpsms wants to pick this up

[joshuaphelpsms]

2/4 - Discussed whether or not the orchestrator code should be moved over. We decided we would just move the python class for citation generator and omit the orchestrator until we feel like we need it. The orchestrator's task is to run experiments and evalauations with various configurations which we do not need for this sample. The citation generator should still accept a question and operate a question at a time. However, we will not be experimenting with different prompts so those can be hardcoded.

@jose-perales, tagging for visability if you wanted to add thoughts here

[jose-perales]

Josh talked with Jeff and decided to keep the orchestrator code, planning on moving other everything for this in the src directory.