Curriculum Vitæ Rahul Jain IBM T J Watson Research, 19 Skyline Drive, Hawthorne, NY 10532

(510) 520-1140 [email protected] http://www.eecs.berkeley.edu/∼rjain

OBJECTIVE

A position in the areas of operations research, decision theory, statistical learning

INTERESTS

Machine Learning and Stochastic Simulation; Game Theory, Auctions and Optimization; Stochastic Processes and Statistics

RESEARCH

Machine Learning: Universal prediction algorithms Statistical Decision Theory: Simulation-based optimization of Markov decision processes

EDUCATION

Ph.D., EECS, University of California, Berkeley (December 2004) Thesis: Efficient Market Mechanisms and Simulation-based Learning for Multi-Agent Systems Committee: Pravin P. Varaiya (Chair), Jean Walrand, V. Anantharam and Jim Pitman M.A., Statistics, University of California, Berkeley (May 2002) M.S., ECE, Rice University (May 1999) B.Tech., EE, Indian Institute of Technology, Kanpur, India (May 1997)

EXPERIENCE

Post-Doctoral Staff Member, IBM Watson Research Center (September 2006 - now) Visiting Scholar, EECS, UC Berkeley (January 2005 - August 2006) Research Assistant, EECS, UC Berkeley (Aug 1999-December 2004) Teaching Assistant, EECS, UC Berkeley (Spring 1998, Spring 1999, Fall 2000, Spring 2004) Visiting Student, AT&T Shannon Labs (August 2001) Research Consultant, Gigabit (later Io-span) Wireless, Inc. San Jose (June-December 1999) Visiting Student, Center for AI and Robotics, Bangalore, India (Summer 1996)

RECOGNITION IBM Invention Achievement Award, 2007 Nomination, The Eli Jury Prize and David Sakrison Best Dissertation Award (2005) Elected Member, Eta Kappa Nu (HKN) (1998) Texas Instruments Fellowship, Rice University (1997-98) TCS Best B.Tech Project Award, IIT Kanpur (1997) Indian National Talent Search Exam (NTSE) Fellowship (1991-97)

AFFILIATIONS Member of IEEE, IEEE Communications, IEEE Control Systems and IEEE Information Theory Society, Member of ACM, INFORMS, SIAM, IMS, AMS, Member of The New York Academy of Sciences

JOURNAL PAPERS

1. R.Jain and P.P.Varaiya, ”Simulation-based uniform value function estimates of Markov decision processes”, SIAM J. Control and Optimization, 45(5):1633-1656, November 2006. 2. R.Jain and P.P.Varaiya, ”Combinatorial exchange mechanisms for efficient bandwidth allocation”, Communications in Information and Systems, 3(4):305-324, September 2004. 3. T.Simsek, R.Jain and P.P.Varaiya, ”Scalar estimation and control with noisy binary observations”, IEEE Trans. Automatic Control, 49(9):1598-1603, September 2004. 4. R. Jain, A.Puri and R.Sengupta, ”Geographical routing for wireless ad hoc networks using partial information”, IEEE Personal Communications, 8(1):48-57, February 2001. 5. R.Jain, ”Key distribution and authentication protocol for secure wireless conferencing”, IETE Technical Review, 15(6), December 1998. 6. R.Jain and P.P.Varaiya, ”The combinatorial sellers’ bid double auction: An asymptotically efficient market mechanism”, under review J. of Economic Theory, last revised, August 2007. 7. R.Jain and J. Walrand, ”An efficient Nash-implementation mechanism for divisible resource allocation”, IEEE J. Selected Areas in Communications, under review, August 2007. 8. P. Dube, R. Jain and C. Touati, “An analysis of pricing competition for queued services with multiple providers”, IEEE J. Selected Areas in Communications, under review, August 2007. 9. R. Jain and P. Varaiya, “Simulation-based optimization of Markov decision processes: An empirical process theory approach”, under preparation, December 2007. 10. R. Jain, “A Design for a bipartite matching market with pair-wise strong budget balance”, under preparation, January 2008. 11. R. Jain and P. Dube, “Pricing games for differentiated queueing services”, under preparation, January 2008.

CONFERENCE PAPERS 1. R. Jain and P. Varaiya, ”PAC bounds for simulation-based optimization of Markov decision processes”, Proc. CDC, December 2007. (invited SIAM paper) 2. R. Jain and P. Varaiya, ”Simulation-based optimization for Markov decision processes and Multi-armed bandits: An empirical process theory approach, Conf. on Stochastic Processes and Applications (SPA-07), Urbana-Champaign, IL, August 2007. 3. R.Jain and P.P.Varaiya, ”Simulation-based uniform estimates of value functions of Markov decision processes”, Conf. Stochastic Processes and Applications, June 2005. 4. R.Jain and P.P.Varaiya, ”Simulation-based uniform value estimates of discounted and average reward MDPs”, Proc. Conf. Decision and Control (CDC), December 2004. 5. R.Jain and P.P.Varaiya, ”PAC learning for Markov decision processes and dynamic games”, Proc. IEEE Symp. on Information Theory (ISIT), June 2004. 6. R.Jain and P.P.Varaiya, ”Extensions to PAC learning for partially observable Markov decision processes”, Proc. Conf. Information Systems and Sciences (CISS), March 2004. 7. T.Simsek, R. Jain and P.P.Varaiya, ”Control under communication constraints”, Proc. Conf. Decision and Control (CDC), December 2002.

8. R.Jain, G.Rajaram, S.Gupta, S.Subramanium, A.Rawat, A.Mukerjee, ”A robot that learns to play soccer”, Proc. Intl. Symp. on Intelligent Robotic Systems, Ed. M. Vidyasagar, January 1998. 9. P. Dube and R. Jain, ”Pricing differentiated queued services”, INFORMS Annual Meeting, Seattle, November 2007. 10. R. Jain, ”Designing a bipartite matching market”, IEEE Conf. on Decision and Control, December 2007. 11. R. Jain, ”Stability of marriage and dowry”, Intl. Conf. on Applied Mathematics, Plovdiv, Bulgaria, August 2007. 12. R.Jain,” A design for an asymptotically efficient combinatorial Bayesian markets: Generalizing the Satterthwaite-Williams mechanism, Int. Conf. on Game Theory, Stony Brook, NY, July 2007. 13. P. Dube, R. Jain and C. Touati, ”An analysis of oligopolistic competitive markets of queued services with QoS guarantees” INFORMS Applied Probability Meeting, July 2007. 14. A.Dimakis, R.Jain and J.Walrand, ”Mechanisms for efficient allocation in divisible capacity networks”, Proc. Conf. Decision and Control, December 2006. 15. C. Kaskiris, Y. Butun and R. Jain, ”An experimental analysis of a combinatorial market mechanism for bandwidth trading”, Proc. of the Workshop on Bandwidth Trading, November 2006. 16. R.Jain and P.P.Varaiya, ”Efficient market mechanisms for network resource allocation”, Proc. Conf. Decision and Control (CDC), December 2005. (invited paper) 17. R.Jain, ”Combinatorial bandwidth exchange: Mechanism and infrastructure”, NSF 100×100 Project Workshop, University of California, Berkeley, March 31-April 1, 2004. (nonarchived) 18. R. Jain, C. Kaskiris, R. Rajagopal and P. Varaiya, ”Combinatorial auction design for bandwidth trading: An experimental study”, accepted International Conf. on Experiments in Economic Sciences (EES), December 2004. 19. R.Jain and P.P.Varaiya, ”An efficient incentive-compatible combinatorial market mechanism”, invited paper, Proc. Allerton Conf., October 2004. 20. R.Jain, A.Dimakis and P.P.Varaiya, ”On the existence of competitive equilibria in bandwidth markets”, Proc. Allerton Conf., October 2002. 21. M.Ergen, S.Coleri, B.Dundar, R.Jain, A.Puri, P.P.Varaiya, ”Application of GPS to mobile IP and routing in wireless networks”, Proc. IEEE Vehicular Tech. Conf., September 2002. 22. R.Jain, ”Geographical routing for wireless ad hoc networks”, IMA Workshop on Hot Topics in Wireless, University of Minnesota, Minneapolis, August 2001. (non-archived) 23. R.Jain, ”Geographical routing for wireless ad hoc networks using partial information”, Workshop on The Foundations of Wireless Networks and Applications, DIMACS Center, Rutgers University and AT&T Shannon Labs, August 2000. (non-archived) 24. R. Jain, B.Sadeghi and E.W.Knightly, ”Towards coarse-grained mobile-QoS”, Proc. WOWMOM, August 1999. 25. R.Jain and E.W.Knightly, ”A framework for design and evaluation of mobile admission control algorithms”, Proc. INFOCOM, March 1999. PROPOSALS

1. ”Multi-agent control: Mechanism design and adaptive learning”, co-authored with Pravin Varaiya, funded by National Science Foundation, August 2004. 2. ”Algorithms and system design for bandwidth trading”, co-authored with Jean Walrand, funded by UC MICRO grant and Fujitsu Labs, May 2005.

PATENTS

1. R.Jain, C.Kaskiris, N.Pillai, R.Rajagopal, J.Shu and P.P.Varaiya, “Method and system for conducting combinatorial auctions”, US Provisional Patent 60/665058, March 23, 2005. (US Patent application file 11/389558), March 23, 2006. (license agreement with Fujitsu, USA) 2. R. Jain, “Method and system for designing bipartite matching markets with applications to online advertising industry”, US patent file 11/837616, filed August 13, 2007. Disclosure no. YOR8-2007-0174 3. R. Jain and P. Dube, “idea Exchange: A method and system for auction of goods with non-exclusivity of use”, approved for filing by Watson Math Dept. patent review board, October 8, 2007. Disclosure no. YOR820070825. 4. P. Dube and R. Jain, “Knowledge Exchange: Method and system for an incentivized reputation system for knowledge trading”, approved for filing by Watson Math Dept. patent review board, October 8, 2007. Disclosure no. YOR820070826. 5. A. Biem, R. Jain and J. Sairamesh, “A system and method to measure the reliability of a new manufacturing process”, under evaluation, August 30, 2007. Disclosure no. YOR8-2007-0495. 6. R. Jain, P. Dube and M. Naphade, “Method for reputation inference across domains in social networks”, under evaluation, January 11, 2007.

PROFESSIONAL SERVICE 1. TPC Member, Workshop on Bandwidth on Demand (BonD 08), April 2008, Brazil. 2. Co-chair, Session on Game Theory, IEEE Conf. on Decision and Control (CDC), December 2007 3. Co-organizer, Session on Pricing for Networks and Queues, INFORMS Annual Meeting, 2007 4. TPC Member, Workshop on Game Theory in Communications (Game-Comm) (A ValueTools Workshop), Nantes, France, October 2007 5. TPC Member, Workshop on Bandwidth on Demand (BonD 06), San Francisco, November 2006. (An IEEE Globecom Workshop) 6. Co-chair, Session on Network pricing, IEEE Conf. on Decision and Control (CDC), December 2006 7. Reviewer for Automatica, IEEE Trans. on Auto. Control, IEEE Trans. on Information Theory, IEEE/ACM Trans. on Networking, IEEE J. Selected Areas in Communications, Queueing Systems, European J. of Control SELECTED GRADUATE COURSE WORK Topology (math 202A), Functional Analysis (math 202B), Algebra, Differentiable manifolds; Probability Theory (stat 205A), Stochastic Processes (stat 205B), Stochastic Calculus (stat 250),Theoretical Statistics (stat 210A), Combinatorial Stochastic Processes (stat 260); Optimization Theory (eecs 227), Information Theory (eecs 290S), Coding Theory (eecs 290S); Statistical Learning Theory (cs 270A), Reinforcement Learning (cs 290), Computation and Complexity Theory (cs 172), Computational and Stochastic Geometry (cs 270B); Nonlinear Systems (eecs 222), Stochastic systems (eecs 223); Communication Networks (eecs 228A), Digital Communication Systems (eecs 224).

TECHNICAL SKILLS

C/C++, Java, Matlab, R, LaTex SELECTED INVITED TALKS 1. ”Simulation-based optimization of Markov decision processes: An empirical process theory approach”, Math Department, IBM Watson Research, Yorktown Heights, November 26, 2008. 2. ”Match-making on demand: From bandwidth to online advertising”, Math Department, IBM Watson Research, Yorktown Heights, May 9, 2007. 3. ”Match-making: From efficient network markets to stable marriages”, EE Department, Yale University, April 11, 2007. 4. ”Economic mechanism design for efficient network operation”, ECE Department, RPI, March 20, 2007. 5. ”Economic mechanisms for efficient network resource allocation”, EE Department, U. Maryland, College Park, February 16, 2007 6. ”Combinatorial market design for efficient resource allocation”, Microsoft Research, Redmond, WA, July 24, 2006. 7. ”Network pricing for efficient resource allocation”, Center for Information Theory and Applications, University of California, San Diego, May 31, 2006. 8. ”Market design and analysis for network systems”, Math Center, Bell Labs, May 10, 2006. 9. ”Simulating Markov decision processes”, ECE Department, University of California, San Diego, March 30, 2006. 10. ”Solving Markov decision processes: Can empirical process theory help?”, Workshop on Learning and Information in Games and Control, Center for Mathematics of Information, Caltech, March 22-23, 2006. 11. ”Efficient market mechanisms for network resource allocation”, EECS Seminar, University of California, Berkeley, November 29, 2004. 12. ”Simulation-based learning for partially observable Markov decision processes and games”, CSPL Seminar, EECS Department, University of Michigan, Ann Arbor, July 6, 2004. 13. ”Combinatorial bandwidth exchange: Mechanism and infrastructure, School of Information and Management Systems, University of California, Berkeley, March 31-April 1, 2004. 14. ”Simulation-based learning for Markov decision processes and games”, Seminar, Mathematical Sciences Research Center, Lucent Technologies Bell Laboratories, March 24, 2004. 15. ”Learning for partially observable Markov decision processes and games”, EE Department Seminar, Cornell University, March 22, 2004. 16. ”Learning for Markov decision processes and games”, EECS Seminar, University of California, Berkeley, March 10, 2004.

REFERENCES

1. Pravin Varaiya, Distinguished Professor EECS Department, 271M Cory Hall, University of California, Berkeley, CA 94720-1770 (510) 642-5270 [email protected] 2. Jean Walrand, Professor EECS Department, 257M Cory Hall, University of California, Berkeley, CA 94720-1770 (510) 219 5821 [email protected] 3. Alexander Kurzhanski, Distinguished Professor Faculty of Computational Mathematics and Cybernetics, Moscow State University. Visiting Professor, EECS Department, 271M Cory Hall, University of California, Berkeley, CA 94720-1770 (510) 526-1525 [email protected], [email protected] 4. Vivek Borkar, Professor and Dean School of Technology and Computer Science, Tata Institute of Fundamental Research, Homi Bhabha Road, Mumbai 400005, INDIA. +91-22-22782293 [email protected] 5. J. Sairamesh, Manager IBM T J Watson Research, 19 Skyline Drive, Hawthorne, NY 10532 (914)-646-3764 [email protected], jramesh [email protected]

CURRENT AND RECENT PROJECTS

1. Risk Management for Complex Supply Chain Networks. (IBM Watson) Today’s complex supply chains of manufacturers depend crucially on third party OEMs. This makes the smooth operation of the supply chain and manufacturing process depend crucially on the OEMs meeting quality specifications, delivery schedules as well as their financial health. These elements introduce additional risk into the supply chain and must be taken into account in designing the right risk metrics. As part of this project, it was my responsibility to define the right risk metrics, computing them using advanced statistical techniques, developing prediction algorithms and overseeing their implementation in a web-based risk management portal. The implementation of the statistical algorithms was done in Matlab and C++ on a windows platform for testing purposes, and in Java for embedding in the portal. The project was led by my manager Dr. J. Sairamesh. 2. Pricing of Queued Services with Strategic Providers. (IBM Watson) Motivated by on-demand IT services, such as call center operations, we studied the pricing problem for providers when other competing providers are present. Moreover, the customers utility is affected by the system load, i.e., the delay. We were able to say what the stable operating points would be for providers in such a competitive environment. The project involved primarily game-theoretic analysis with some Matlab simulation on windows. This is ongoing work jointly with Parijat Dube, IBM Watson. 3. AdExchange: Design for an Online Advertisement Exchange. (IBM Watson) The success of Google, Yahoo and others in provision of online advertising services has led to new models and problems to further expand this market. This project proposes design for an advertisement exchange that matches advertisers with hosts. This requires bids from both sides. The mechanism then determines the optimal matches and the payments, which are designed to make the system robust to manipulation by any of the participants. Further, the mechanism is pair-wise strongly budget-balanced, ex ante individual rational and an efficient Nash-implementation under certain assumptions.