I am a Senior Research Scientist at Google specializing in designing and building large-scale machine learning systems and models for personalization, content analysis and user modeling. I received my M.Sc and PhD degrees from the School of Computer Science, Carnegie Mellon University in 2009 and 2011, respectively. I became a Google Research Scientist in Mountain View in 2012 after spending a year at Yahoo! Research in Santa Clara. I am the recipient of the 2012 ACM SIGKDD Doctoral Dissertation Award in addition to several best paper awards at KDD 2014, WSDM 2014, and WSDM 2012 (runner up). Specialties: Machine learning/Data mining, Latent variable models, Web-scale distributed inference algorithms, User modeling, Personalization, Recommender systems, Social Media, content analysis, and computational advertising. I published over 50 papers and co-presented 5 tutorials at WWW'11,NIPS'11, WWW'12, KDD'13 and AAAI'14 on these topics. I served as an Area Chair for ICML 2014, ICDM 2014 and WSDM 2015. See more details at : research.google.com/pubs/AmrAhmed.html‎

Senior Research Scientist (Technical Lead) @ * Scalable Machine learning Framework (Tech-Lead): building and designing a machine learning platform and a programming model for effortlessly scaling machine learning applications to web-scale data. --> A Patent was filed out of this project. * Understanding User location and geo-targeting. --> A patent was filed out of this project. * Scalable and flexible product recommendation engine. * R&D: various projects exploring novel directions in unsupervised information extraction and targeting (see research.google.com/pubs/AmrAhmed.html‎ for more up-to-date coverage) From August 2012 to Present (3 years 5 months) Research Scientist @ Personalization, product recommendation, and content analysis. Building web-scale distributed machine learning systems for online targeting and user profiling. Main Projects: • Modeling time-varying user interests: How to learn and detect user intents from browsing behaviors? How to leverage this representation for advertising? How to update the intent representation as time goes by? --> Impact: this system is currently in production as part of Yahoo! advertising stack. • Modeling user purchase behavior for product recommendation. Incorporated a model of time, item attributes and product taxonomy into collaborative filtering models. --> Impact: improved the ROC measure by 7-10% over state-of-the-art models. • Large-scale machine learning: How to build infrastructure that supports inference over hundred of millons of user? ( This work was awarded a runner-up best paper award at WSDM 2012 -- Scalable Inference in Latent Variable Models). --> Impact: this work powers a lot of the latent variable models projects/systems at Yahoo! . • Learning over web-scale social and natural graphs: How to leverage social data on the web? How to perform factorization over web-scale graphs? --> Impact: built a model and an inference algorithm that scales to that Y!M graph that has 100s of MILLIONS of users and 10s of BILLION of edges. • Personalized news recommendation: How to model user preferences for news recommendation using implicit feedback? How to personalize story recommendation using the learnt model? (see paper at WSDM 2012 -- Fair and Balanced: Learning to Present News Stories) --> Impact: this work improved CTR by 15% inside Yahoo! • Modeling user geographic location from tweets: How to predict user location from his/her tweets? ( see paper at WWW 2012 -- Discovering Geographical Topics from Twitter Streams) --> Impact: this work achieves state of the art results for location prediction. • Tera-scale multi-task feature selection for advertising From 2011 to October 2012 (1 year) Graduate Research Assistant @ My thesis title was "Modeling Users and Content: Structured Probabilistic Representation, and Scalable Online Inference Algorithms". The thesis covers techniques and algorithms for building models form streaming data at web-scale. Application areas incudes: content understanding, modeling ideological perspective in blogs, building a structured representation of streaming news, bio-text, social network analysis, and user profiling. From 2005 to December 2010 (5 years) Research Scientist Intern @ Dynamic construction of user profiles from streaming data. From June 2010 to September 2010 (4 months) Research Intern @ Multi-task learning in feed-forward deep architectures. From June 2007 to September 2007 (4 months)

PhD, Computer Science @ Carnegie Mellon University From 2005 to 2011 Computer Science @ University of Illinois at Urbana-Champaign Amr Ahmed is skilled in: Machine Learning, Scalable Inference, Statistical Modeling, Data Mining, Content Analysis, User Modeling, Personalization, Computational Advertising, Social Media, Scalable Non-Parametric Bayesian Models, C++, Hadoop, MapReduce, Recommender Systems, Non-parametric Bayesian Models