I like working on problems where the business already has language.
Elasticity is one of those. People who have never seen a notebook still know that some things cannot take a price increase. That shared vocabulary is rare. It is also dangerous, because it makes a bad estimate sound like common sense.
The commercial object
A pricing system does not sell a model. It sells a feasible price: margin, stock, season, and a reason a human can reject. Elasticity is an input to that object, not the object.
That is why I keep putting inventory next to the curve. A highly elastic title with two weeks of stock is a different decision than a staple with a year of supply. If your optimizer does not know that, it is maximizing a homework problem.
What I tell students
If you want to work in applied ML, learn one concept that already lives in the P&L. Elasticity, hazard, lift, elasticity's cousin "incrementality." Then learn why the naive estimator is wrong. Then learn how to ship the less-wrong one.
That path is slower than another fine-tune. It is also how you become the person in the room who can connect a coefficient to a decision.
If you want help walking that path, start from the thing that's stuck. If you want to work on a pricing or measurement problem, write to me.