Uplift modeling: deciding who to target with a marketing campaign to maximize incremental revenue; not just who buys. Experiment design, S/T/X-learners, Qini evaluation, and a cost aware targeting policy with a Streamlit app.

Causal inference for promotional targeting: who should receive the email? Five CATE estimators evaluated by Qini & SNIPS policy value on Hillstrom 2008.

Causal uplift modeling, predict who is persuadable, not who converts. S/T/X/R meta-learners, Qini eval, FastAPI decision API. Hillstrom + Criteo adapters.