Lies, Damn Lies, and Statistical Learning

christopher steger, ’08, netflix

While “lies” is hyperbolic, there are many misconceptions about the people, the problems, and the practical considerations involved with building the machine learning algorithms that power modern consumer products. Using examples from my time at Netflix, I will offer some insights into who builds the recommendation systems, how they do it, and the indispensable skills that are keys to the job.

About Christopher:

Under the tutelage of Behnaam Aazhang and Ashu Sabharwal, Chris concluded his protracted Rice career by completing his Ph.D. in 2008. Upon graduation into a global financial crisis, he found the market for asymptotic bounds on communication over two-way wireless channels to be fairly soft, so instead he joined a startup facing some interesting estimation and detection problems. At Skyhook Wireless, he helped to develop algorithms for location estimation and organic bootstrapping of a positioning system based on wi-fi access points. Then, after the wi-fi positioning market was commoditized by the entry of a succession of technology giants, Chris pivoted to product management to develop a suite of products for mining mobile location data. From there, he made the move to Netflix, a streaming video entertainment company, where he is currently Director of Product Innovation for Personalization Systems, and he leads development of Netflix search and homepage algorithms..