I am currently working on a project that is using AI and ML to add new and powerful enrichments to unstructured data piles.
I just got to integrate Azure Python Functions for interactions between the system and our AI models. It worked super well and here are some of the highlights:
- We now can sub in and out AI models without a rebuild or recompile of the system. -The code is integrated with our main codebase and is deployed through a CI/CD pipeline. (No running the model on a laptop)
- The data scientists on the team can generate the models and just upload them. No need to worry about integration with the product, for them, it just works!
- Azure functions allow us to scale with the needs of the system vs allocating large VMs for AI work that sit unused for long periods.
- This allows us to integrate the AI work into the system without having to bring in a container system just for the models.
Check it out: