Faculty Recruiting Support CICS

Driving Network Growth at LinkedIn: Past, Present and Future

20 Sep
Friday, 09/20/2019 1:30pm to 2:30pm
Computer Science Building, Room 151
Seminar
Speaker: Hema Raghavan, Director of Engineering, LinkedIn

At LinkedIn our mission is to use AI to connect every member of the global workforce to make them more productive and successful. The social network is the backbone for professionals to engage with each other at every stage of their career. LinkedIn's "People You May Know" (PYMK) product has fueled viral growth to connect the world's professionals to each other to form an active community. PYMK has also driven the innovation in technologies and platforms like Kafka and Voldemort/Venice. More recently LinkedIn's PYMK recommendations have moved to a near real time graph walk platform - GAIA. The talk will showcase how the trinity of Artificial Intelligence, Big Data Platforms and Product Innovation are closely coupled.

 Hema Raghavan is a Director of Engineering at Linkedin currently heading the team that builds AI solutions for fueling the professional social network's growth. At LinkedIn she has led the "People You May Know" team through a multi-year transformation of near real time computation. Her team also builds the decision making engine for LinkedIn's notification. She has several years of work experience in Applied AI settings in industry and academia solving hard problems and leading teams to significant impact. Prior to LinkedIn, she was a Research Staff Member at IBM T.J Watson building question answering systems for the DARPA GALE project. She started her career in the industry in Yahoo Labs as a research scientist where she transformed the retrieval engine for Search Advertising leading to significant revenue lifts. She received her PhD from the University of Massachusetts, Amherst in Information Retrieval and Machine Learning. She has several research publications in top-tier conferences like WWW, SIGIR, etc. She serves on the program committee for several conferences like WWW, KDD and others.