E RIC S ODOMKA
*** ******** ******, *** *****, Providence, RI 02912 406-***-****; abpi4t@r.postjobfree.com
Objective
I am a Ph.D. Candidate in Computer Science interested in solving current research
problems in electronic markets. I am most interested in modeling buyer and seller behavior,
better understanding auction dynamics, providing decision support for market
participants, or any similar challenges in online auctions or other strategic domains.
Education
Brown University Providence, RI
Ph.D. Candidate, Computer Science, 2007-present
Sc.M., Computer Science, 2009
Cumulative GPA: 4.0/4.0
Advisor: Amy Greenwald
Relevant Coursework: Design of Agents for Bidding in Sponsored Search Auctions,
Game Theory, Introduction to Combinatorial Optimization, Probabilistic Methods in
Computer Science, Game-Theoretic Arti cial Intelligence, Solving Hard Problems in
Combinatorial Optimization, Recent Applications of Probability and Statistics,
Approximation Algorithms
University of Minnesota Minneapolis, MN
B.S. Computer Science, 2006
Graduated Summa Cum Laude.
Advisor: Maria Gini
Experience
Google New York, NY
Software Engineering Intern: DoubleClick for Publishers, Summer 2012
Implemented dynamic programming, linear programming, and controller-based
methods of optimizing publisher revenue given Ad Exchange (AdX) bid distributions.
Created prediction and optimization testing frameworks for rapid prototyping and
evaluation of price predictors and optimization algorithms.
Visualized publisher data on various dimensions, such as impression competition
amongst a publisher s advertisers, expected revenue over the space of possible actions,
and revenue function approximations with respect to advertiser contract satisfaction.
Yahoo! Research New York, NY
Research Intern: Economics and Social Systems, Fall 2010
Implemented methods to infer advertiser values-per-click based on bidding behavior
across keywords.
Created hierarchical regression model to (1) understand which keyword features were
driving each advertiser s bids and values-per-click, and (2) predict advertiser
bids/values on new keywords.
Constructed models of advertiser click/cost beliefs from sponsored search logs.
The Search Agency East Greenwich, RI
Research Intern, 2009-2010
Developed methods to estimate sponsored search auction parameters using Google
bid simulator data from multiple accounts.
Created simulations to better understand counter-intuitive auction dynamics with
budget-constrained bidders, such as potential non-monotonicity in clicks with
increasing bids.
Brown University Providence, RI
Graduate Teaching Assistant, 2009-2011
Courses TAed: Autonomous Agents and Computational Market Design, Design of
Agents for Bidding in Sponsored Search Auctions, Introduction to Arti cial
Intelligence.
Co-led students in designing various components of autonomous sponsored search
bidding agents which ran in the Trading Agent Competition for Ad Auctions.
Led discussions on recent publications in sponsored search research.
Select Awards
Yahoo! Key Scienti c Challenges Award Recipient: Computational Advertising, 2012.
Facebook Fellowship Finalist, 2012.
TAC AA Finalist, 2009, 2010, 2011, 2012.
TAC SCM Finalist, 2005, 2006, 2008, 2009, 2010.
TAC Best Poster Award, May 2006.
Research Projects
Dutch Flower Auctions Rotterdam, The Netherlands
January 2011-present
Formulated MDP-based best-response algorithms to search for epsilon-equilibria in a
stylized model of the Dutch Flower Auctions.
Developed a simulator based on actual Dutch Flower Auction data to enable
evaluation of various price predictors and bidding strategies.
Trading Agent Competition for Ad Auctions Brown University
2008-present
Achieved 2nd place in the annual Trading Agent Competition for Ad Auctions in
2009, 2010, 2011, and 2012.
Developed demand predictors and multiple-choice knapsack-based optimization
algorithms for budget-constrained bidding in sponsored search.
Formulated mathematical programs to disaggregate average position data to more
accurately estimate the number of impressions seen by each advertiser in each slot.
Trading Agent Competition for Supply Chain Management Brown University, Univer-
sity of Minnesota
2004-2010
Reached nals of the annual Trading Agent Competition for Supply Chain
Management in 2005, 2006, 2008, 2009, and 2010.
Led Brown University s agent in 2008-2010, and team member for the University of
Minnesota s agent in 2004-2007.
Designed and implemented various supplier price predictors, sales price predictors,
procurement algorithms, and bidding algorithms using both mathematical
programming techniques and greedy marginal pro t-based algorithms.
Created a controlled testing framework to assist in game analysis, sensitivity analysis,
and variance reduction.
Publications
Amy Greenwald, Jiacui Li, and Eric Sodomka. Approximating Equilibria in Sequential
Auctions with Incomplete Information and Multi-Unit Demand. Neural Information
Processing Systems (NIPS), December 2012.
Eric Sodomka, Sebastien Lahaie, and Dustin Hilliard. A Hierarchical Model for Value
Estimation in Sponsored Search. INFORMS 2012 Annual Meeting, October 2012.
Michael P. Wellman, Eric Sodomka, and Amy Greenwald. Self-Con rming Price Prediction
Strategies for Simultaneous One-Shot Auctions. Twenty-Eighth Conference on Uncertainty
in Arti cial Intelligence (UAI), August 2012.
Amy Greenwald, Jiacui Li, and Eric Sodomka. Approximating Equilibria in Sequential
Auctions with Incomplete Information and Multi-Unit Demand. ICML-12 Workshop on
Markets, Mechanisms, and Multi-Agent Models, July 2012.
Thomas Goff, Amy Greenwald, Elizabeth Hilliard, Wolfgang Ketter and Eric Sodomka.
JACK: A Java Auction Con guration Kit. AAMAS-12 Workshop on Agent-Mediated
Electronic Commerce (AMEC) and Trading Agent Design and Analysis (TADA), June 2012.
Michael Wellman, Eric Sodomka, and Amy Greenwald. Price Prediction Strategies for
Simultaneous One-Shot Auctions. In NSF/CEME Decentralization Conference. March
2012.
Eric Sodomka, Jordan Berg, Amy Greenwald, Wolf Ketter, Yixin Lu, and Eric van Heck.
Autonomous Trading in Simulated Dutch Flower Auctions. INFORMS 2011 Annual
Meeting. November 2011.
Jordan Berg, Carleton Coffrin, Amy Greenwald and Eric Sodomka. Rank and Impression
Estimation in TAC AA. In IJCAI 2011 Workshop on Trading Agent Design and Analysis.
July 2011.
Eric Sodomka, Sebastien Lahaie, and Dustin Hillard. A Hierarchical Model for Value
Estimation in Sponsored Search. In EC 2011 Seventh Ad Auctions Workshop. June 2011.
Jordan Berg, Amy Greenwald, Victor Naroditskiy, and Eric Sodomka. A Knapsack-Based
Approach to Bidding in Ad Auctions. In the 19th European Conference on Arti cial
Intelligence (ECAI). August 2010.
Jordan Berg, Amy Greenwald, Victor Naroditskiy, and Eric Sodomka. A First Approach to
Autonomous Bidding in Ad Auctions. In EC 10 Workshop on Trading Agent Design and
Analysis (TADA 2010). June 2010.
Amy Greenwald, Victor Naroditskiy, Tyler Odean, Mauricio Ramirez, Eric Sodomka, Joe
Zimmerman, and Clark Cutler. Marginal Bidding: An Application of the Equimarginal
Principle to Bidding in TAC SCM. In edited volume on Agent Mediated Electronic
Commerce and Trading Agent Design and Analysis, Lecture Notes in Business Information
Processing, Volume 13. 2009.
Eric Sodomka, John Collins, and Maria Gini. Ef cient Statistical Methods for Evaluating
Trading Agent Performance. Proceedings of AAAI-07, Vancouver, Canada, July 2007.
Wolfgang Ketter, Eric Sodomka, Amrudin Agovic, John Collins, and Maria Gini. Strategic
Sales Management in an Autonomous Trading Agent for TAC SCM (demonstration).
Proceedings of AAAI-06, Boston, Massachusetts, USA, July 2006.
Brett Borghetti and Eric Sodomka. Performance Evaluation Methods for the Trading Agent
Competition (student abstract). Proceedings of AAAI-06, pp 1857- 1858, Boston,
Massachusetts, USA, July 2006.
Brett Borghetti, Eric Sodomka, John Collins, and Maria Gini. A Market-Pressure-Based
Performance Evaluator for TAC-SCM (poster). Proceedings of AAMAS/ TADA-AMEC-06,
Hakodate, Japan, May 2006.
Paul Schrater, Eric Sodomka, Charles Sloane. Decision-Making with Monetary Value
Uncertainty (poster). Proceedings of the Society for Judgment and Decision Making
Conference, Toronto, Canada, November 2005.