David Sweet has spent his career tuning the kinds of systems where small improvements are worth a great deal. As a quantitative trader at GETCO, he used experimental methods to tune trading systems, where decisions happen in microseconds and evidence beats intuition. As a machine learning engineer at Instagram, he applied the same experimental mindset to recommender systems at consumer scale.

The common thread in David’s work is treating system improvement as a measurement problem. Rather than shipping changes on instinct, he adjusts a system, observes the response, and uses that direct feedback to decide what to keep. It is a discipline drawn from control theory and experimentation, and it applies equally to trading infrastructure and to the products millions of people use every day.

David has also become a teacher of these methods. His book grew out of the lectures on tuning quantitative trading systems that he has given at NYU Stern over the past three years, extending his classroom material into a form other engineers can practice with. Through his teaching and writing, he helps practitioners replace hunches with experiments when they optimize the systems they run.