Olya Ohrimenko @ UniMelb

Home Research Service About

I joined School of Computing and Information Systems as a senior lecturer in early 2020. Prior to joining the University of Melbourne I spent 6 years at Microsoft Research in Cambridge, UK, where I was a principal researcher leading work on Confidential AI and I/O side-channel mitigation. Between 2014 and 2016, I was the Microsoft Fellow at Darwin College in Cambridge University.

My research interests include privacy, integrity and security of machine learning algorithms, data analysis tools, multi-party computation and systems relying on cloud storage, computation and hardware. The goal is to identify and prevent potential vulnerabilities and to design and build efficient solutions with provable guarantees. I also enjoy working on algorithms, data structures and theory.

If you are interested in pursuing a Ph.D. or a Master's project in any of the areas listed above, please contact me directly.

Contact: oohrimenko @ unimelb dоt edu dоt au

News:

  • Organising workshop on Privacy-Preserving Machine Learning at NeurIPS 2020.
  • Received Facebook research award on the "Role of applied cryptography in a privacy-focused ecosystem".
  • Received Facebook "Probability and Programming research award" with Ben Rubinstein and Toby Murray.
  • Our paper on Analyzing Information Leakage of Updates to Natural Language Models with Santiago Zanella-Béguelin, Lukas Wutschitz, Shruti Tople, Victor Rühle, Andrew Paverd, Boris Köpf, and Marc Brockschmidt to appear at CCS 2020.

Areas I have worked on include:

  • multi-party machine learning: confidentiality, privacy, data contamination, incentives
  • data-oblivious algorithms for memory access, shuffling and sampling data, MapReduce, machine learning and graph drawing
  • differential privacy
  • cache and memory side-channels in secure hardware (e.g., Intel SGX)
  • verifiable computation and authenticated data structures (with zero knowledge)
  • searchable encryption
  • blockchain framework and applications
  • constraint programming and optimization

Recent publications

(Full list here)

  • Analyzing Information Leakage of Updates to Natural Language Models by S. Zanella-Béguelin, L. Wutschitz, S. Tople, V. Rühle, A. Paverd, O. Ohrimenko, B. Köpf, and M Brockschmidt. In ACM Conference on Computer and Communications Security (CCS) 2020.
  • Oblivious Sampling Algorithms for Private Data Analysis by S. Sasy and O. Ohrimenko. In NeurIPS, 2019.
  • An Algorithmic Framework For Differentially Private Data Analysis on Trusted Processors by J. Allen, B. Ding, J. Kulkarni, H. Nori, O. Ohrimenko, and S. Yekhanin. In NeurIPS, 2019.
  • Contamination Attacks and Mitigation in Multi-Party Machine Learning by J. Hayes and O. Ohrimenko. In NeurIPS, 2018.
  • Structured Encryption and Leakage Suppression by S. Kamara, T. Moataz, and O. Ohrimenko. In CRYPTO, 2018.

@Microsoft

At Microsoft Research I was part of Confidential Computing group when our group contributed to:

  • Azure Confidential Computing
  • Projects highlighted by CTO of Microsoft Azure: oblivious computation, multi-party ML, and confidential inference (1:06:08).
  • Confidential Consortium Framework

I was also lucky to work with many bright interns:

  • Dongge Han (Oxford University), 2019
  • Wanrong Zhang (Georgia Institute of Technology), 2019
  • Sajin Sasy (University of Waterloo), 2019
  • Jamie Hayes (University College London), 2018
  • Kartik Nayak (The University Of Maryland), 2017
  • Lawrence Esswood (Cambridge University), 2016
  • Daniel Gruss (Graz University of Technology), 2016
  • Sameer Wagh (Princeton University), 2016
  • Esha Ghosh (Brown University), 2015
  • Aastha Mehta (MPI-SWS), 2014
  • Divya Sharma (Carnegie Mellon University), 2014

Contributed to the report on Privacy Enhancing Technologies by the Royal Society.

(Some) Invited talks

  • Secure hardware for privacy-enhanced computation, UK Royal Society, 2019
  • Update on Confidential Computing, RSA, 2019
  • Confidential Machine Learning on Trusted Processors, Data, Learning and Inference (DALI), 2018
  • Sailing across the Side channel towards Confidential Computing, Real World Crypto, 2018
  • Confidential Machine Learning, AI Summer School, 2017

Co-organized workshops

  • Privacy Preserving Machine Learning, CCS 2019
  • Privacy Preserving Machine Learning, NeurIPS 2018

Program committees

  • CRYPTO 2020
  • Computer Security Foundations Symposium (CSF 2020)
  • Real World Crypto (RWC 2020)
  • ICML workshop on Privacy (PiMLAI 2018)
  • Conference on Computer and Communications Security (CCS 2018, CCS 2019)
  • Financial Cryptography and Data Security (FC 2018, FC 2019)
  • Conference on Distributed Computing Systems (ICDCS 2016)
  • CCS Workshop on Privacy in the Electronic Society (WPES 2014, WPES 2015)

Reviewer for ICML 2020, AAAI 2020, NeurIPS 2019, ICML 2019, NeurIPS 2018.

Education

  • Sc.M., Ph.D. Brown University
  • B.CS. (Hons) The University of Melbourne

Internships

  • 2012 IBM Research Zurich with Christian Cachin
  • 2012 MSR Redmond with Seny Kamara
  • 2010 Google NYC, Search team with Jeff Cox
  • 2009 Google CA, AdSense
  • 2008 Optiver, Australia

Other

  • 17 for ’17: Microsoft researchers on what to expect in 2017 and 2027
  • Postdoc attendee of Heidelberg Laureate Forum
  • On judge panel of Codess Hackathon
  • My Erdős number is 3 (times 3 from Claire Mathieu, Michael Mitzenmacher and Eli Upfal)
  • Olya vs. Olga: Both are fine with a slight preference for the former