Pinar Ozisik

Ph.D. in Computer Science
College of Information and Computer Sciences
University of Massachusetts Amherst


I am currently in the job market, looking for a position in industry or academia (mainly postdocs). Please feel free to email me if you think I'd be a good fit for your group. Here is a copy of my CV.

I completed my Ph.D. in UMass Amherst, where I was a member of the Cryptoeconomics Lab, working with Prof. Brian Levine, and was co-affiliated with the Autonomous Learning Lab, collaborating with Prof. Philip Thomas. My thesis focused on the analysis and application of concentration inequalities (also known as tail inequalities) to preserve the desirable properties of a given system. More specifically, I worked on a protocol that used tail inequalities to improve the network performance of blockchain systems; and I analyzed the susceptibility of a class of reinforcement learning algorithms that leveraged tail inequalities to data poisoning attacks.

Conference and Workshop Papers

  • Security Analysis of Safe and Seldonian Reinforcement Learning Algorithms. [supplemental] [video] [code] [poster]
    A. Pinar Ozisik, and Philip S. Thomas. In Neural Information Processing Systems (NeurIPS), December 2020. (20.1% acceptance rate)
  • Graphene: Efficient Interactive Set Reconciliation Applied to Blockchain Propagation. [video] [code]
    A. Pinar Ozisik, Brian N. Levine, George Bissias, Gavin Andresen, Darren Tapp, and Sunny Katkuri. In Conference of the ACM Special Interest Group on Data Communication (SIGCOMM), August 2019. (14.5% acceptance rate)
  • Graphene: A New Protocol for Block Propagation Using Set Reconciliation. [short version]
    A. Pinar Ozisik, Gavin Andresen, George Bissias, Amir Houmansadr, and Brian N. Levine. In ESORICS International Workshop on Cryptocurrencies and Blockchain Technology (CBT), September 2017.
  • Sybil-Resistant Mixing for Bitcoin.
    George Bissias, A. Pinar Ozisik, Brian N. Levine, and Marc Liberatore. In Proceedings of ACM Workshop on Privacy in the Electronic Society (WPES), November 2014.
  • Detecting Stumbles with a Single Accelerometer.
    Nabil Hajj Chehade, A. Pinar Ozisik, James Gomez, Fabio Ramos, and Gregory J. Pottie. In International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), August 2012.
  • The Effects of Finite Populations and Selection on the Emergence of Signaling.
    Kyle I. Harrington, A. Pinar Ozisik, and Jordan Pollack. In Proceedings of Artificial Life (ALIFE) XIII, July 2012.
  • The Effect of Tags on the Evolution of Honest Signaling.
    A. Pinar Ozisik and Kyle I. Harrington. In Proceedings of the Genetic and Evolutionary Computation Conference (GECCO) Companion, July 2012.
  • Other Write-ups

  • Concentration Inequalities in the Wild: Case Studies in Blockchain & Reinforcement Learning.
    A. Pinar Ozisik. Doctoral Dissertation, February 2021.
  • Estimation of Miner Hash Rates and Consensus on Blockchains.
    A. Pinar Ozisik, George Bissias, and Brian N. Levine. arXiv preprint arXiv:1707.00082, July 2017.
  • An Explanation of Nakamoto's Analysis of Double-spend Attacks.
    A. Pinar Ozisik, and Brian N. Levine. arXiv preprint arXiv:1701.03977, January 2017.
  • An Analysis of Attacks on Blockchain Consensus.
    George Bissias, Brian N. Levine, A. Pinar Ozisik, Gavin Andresen, and Amir Houmansadr. arXiv preprint arXiv:1610.07985, October 2016.