Deploying ML models to production. I work in silicon at an established company, and management every year wants to incorporate “ML” and “AI” into our workflows

Deploying ML models to production. I work in silicon at an established company, and management every year wants to incorporate “ML” and “AI” into our workflows, but it doesn’t happen because a) nobody wants to risk their review on it b) the median age is 40 and it requires a brain rewiring. I spent part of a bout of FMLA learning to develop simple PyTorch models to help with our data processing, which is sometimes frustratingly qualitative because nobody can define rigorously what they want to compute, just endless sketches and corner cases