You may need to have your tables in Azure SQL Database using the relationships (don't forget the primary and foreign keys). You may also need to consider creating stored procedures or views to aggregate or filter data for specific tasks.
Then, you need Azure Cognitive Search to index data from multiple tables in your Azure SQL Database.
In reality, the indexer in Azure Cognitive Search can traverse the relationships in your tables and create a search index that respects those relationships. This can be useful if you want to do a free-text search over your data.
Azure Machine Learning Service provides cloud-based tools for building and deploying machine learning models. You can use this service to build custom models on top of your data and if you have a pre-trained OpenAI model, you can deploy it on Azure using Azure ML Service or Azure Kubernetes Service (AKS).
Check the following documentation : https://learn.microsoft.com/en-us/azure/ai-services/openai/overview