Nemanja Petrovic
Web: http://www.ifp.uiuc.edu/~nemanja

Interest

I like to understand, build and run scalable, complex systems with a high practical value.

Educational
Background

Ph.D. in Computer Engineering, 2004, University of Illinois M.S. in Computer Engineering, 2001, University of Illinois B.S. in Electrical Engineering, 1999, University of Belgrade

Research and
Working
Experience

Real-time Auctions/Machine Learning/Image Processing/Computer Vision: Google Inc, New York City, NY. Engineer/Technical lead/People manager (2005-present) Initiated, led, executed and productionized various efforts in some of the major focus areas for Google. Currently, I am dual-role technical lead/manager in YouTube Live, responsible for discovery and search of Live streams on all YouTube/Google platforms. Prior to that, I was founding engineer in DoubleClick Bid Manager (DBM) in New York office. DBM (bidder and by side platform) connects world largest trading desks, agencies and advertisers with consolidated supply from 50+ exchanges, processing millions of transactions in a second with 30ms latency. I helped grew video and mobile businesses from zero to $1B figure in under 30 months, and managed 18 engineers. Previously, I was an architect and implementer of a generic nearest neighbor search engine that powers “similarity search” in several Google properties with tens of billions of data points. This engine includes an array of state of the art hashing and partitioning algorithms and high performance architecture and serving stack for approximate nearest neighbor search (k-NN). Prior to that, I was responsible for driving and improving AdWords Budget Optimizer. Even earlier, I wrote probably the first method for large scale distributive blending of language models used for speech recognition and machine translation, and I was a technical lead of visual product search app for mobile phones. Launches include new products and features in Display Ads, Image Search, Voice Search, YouTube and Android. My core technical work is on par with other half of work I am doing: finding gaps in thinking, creating rough ideas, driving them and seeing them through. Siemens Corporate Research, Princeton, NJ. Member of technical staff (2004-2005) Introduced a version of boosting for learning in the presence of missing data. University of Illinois Urbana, IL. Research Assistant (1999-2004) Coursework & research towards MS/PhD theses. Published 24 papers. Internships Microsoft Research, Redmond, WA (three times) University of Waterloo, Waterloo, Ontario, Canada California Institute of Technology, Pasadena, CA University of Illinois, Urbana, IL Deutsche Telekom AG, Darmstadt, Germany Technical University Athens, Greece

Technical
skills

Core knowledge C++, Google technology stack, Cloud technology stack Other technologies Python, Bash, Perforce, AJAX, SWIG, Matlab, VAST and MPEG standards

Ph.D.
Coursework

Communications II, Random Processes, Fundamentals of Speech Processing and Recognition, Control System Theory and Design, Electromagnetic Waves and Radiating Systems, Computer Vision, Topics in Communications: Advanced Coding Theory, Advanced Digital Signal Processing, Learning Algorithms in Signal Processing, Mathematical Statistics II, Advanced Geometric Approach to Computer Vision and Vision Based Control, Applied Stochastic Processes, Business Plan Workshop for Technology Entrepreneurs

Patents

Several patents awarded during my tenures at Microsot, Siemens and Google.

Awards

NSF travel grant, 2001 Highest GPA graduate in 1999, Electrical Engineering, University of Belgrade Summer Undergraduate Research Fellowship, California Institute of Technology, Pasadena, 1998 John Nuveen Stipend, University of Illinois, 1998 Excellent Student Scholarship, Serbian Academy of Sciences and Arts, 1998 DAAD - German Government Academic Exchange Service Fellowship, 1997

Grants

Wrote technical proposal: ARDA/NSA VACE Phase II contract MDA904-03-C-1787 ($641,677.00 for 2 years), PI Thomas S. Huang, 2003 Wrote technical proposal: Microsoft Corp. equipment grant, PI Thomas S. Huang, 2003

Professional
Activities

Member of the Program Committee: ICCV 2007 CVPR 2007 CVPR 2007 Workshop "Beyond Patches" CVPR 2007 Workshop on Semantic Learning Applications in Multimedia ICCV 2007 Workshop on Dynamical Vision CVPR 2006 ECCV 2006 Workshop on Dynamical Vision CVPR 2006 Workshop "Beyond Patches" CVPR 2006 Workshop on Semantic Learning Applications in Multimedia ICCV 2005 Workshop on Dynamical Vision Reviewer: IEEE Trans. Pattern Analysis and Machine Intelligence IEEE Trans. Circuits and Systems for Video Technology IEEE Trans. Information Theory ICCV CVPR BMVC ACM Multimedia MICCAI EURASIP Graphical Models MTAP JIFS IMP ICPR ICIP ICME VLBV