Have you ever wondered what’s normal when it comes to model governance requests from banks? I recently found myself in a pickle, trying to navigate the never-ending list of questions from our banking clients. As someone who works at a company providing inference as a service, I’ve come to realize that the amount of information they request is staggering.
From specific features used in the model to detailed explanations of our modeling methodology, the questions seem endless. And let’s not forget the time frame for train/test/val sets – down to the day! It’s no wonder that our privacy team is overwhelmed, and I’m left wondering if we’re keeping things too close to the chest.
I’m not alone in this struggle. I’ve spoken to others in the industry, and it seems that banks are notorious for their rigorous model governance requests. But what’s normal, and what’s excessive? Are there industry standards that we should be following?
As I dug deeper, I realized that there’s a lack of resources out there showing what’s industry standard. It’s a gray area, and one that requires a delicate balance between transparency and confidentiality. So, I’m turning to you – how does your organization handle model governance requests? Are there any resources you can share that might help us improve our reporting to these concerned customers?
Let’s work together to shed some light on this conundrum and find a way to make model governance more efficient and effective.