With the advancement of AI, it seems there isn’t a task unsolvable by this amazing technology, and no wonder every type of product wants to leverage it. You hire the most brilliant researchers and wait for the magic to happen, but apparently, not every scientist can translate a business problem into a research task, and not every business leader can vividly define achievable goals for the ML organization.

Imagine a digital health company that holds electronic medical records of thousands of patients. I believe we would all immediately assume their ML organization is creating some form of a diagnostics model, to ultimately improve the patients’ health. But, is that necessarily the best way to improve someone's health given this type of data? Maybe it would be more helpful to predict the severity of each patient's case and offer prioritization? And who said that improving patients’ health was even the company's objective? Maybe their objective is to optimize the doctor’s clinic operation.

Therefore, it is key to master the art of understanding the business needs and translating them to an AI research project. In this talk I will share my technique for learning a new business from a data science point of view and provide the pillars for building a realistic roadmap for an ML organization in a way that optimizes the business’ needs.