Biography

I am a technology leader, applied scientist, cryptographer, focusing on real-world security & privacy problems and its intersection to AI/ML. I build and lead high-performing teams working on problems that impact this world. I maintain my academic interests, but I am more passionate about bringing advanced tech to real problems, not hypothetical ones.

I'm generally interested in the problem of cryptographic computation. I'm also recently interested in applying cryptography to the blockchain & cryptocurrencies. I obtained my Ph.D. in Computer Science at Boston University. My past research has generally focused on developing privacy-enhancing technologies that minimize the information being revealed when outsourcing massive datasets in cloud-based environments. (Oh, I sometimes go by Jery :)

News: I'm building a new thing! Stay tuned!

My Work & Research

My Erdős number is 3. I investigate how humans can benefit from technological advancements and how we can let cooler and more advanced technologies gain massive adoption. I used to and still am exploring and researching privacy-enhancing learning technologies, cryptographic protocols, etc.

My blog post on Amazon.science Machine learning models that act on encrypted data talks about a joint work with Prof. Joan Feigenbaum. Open-source code can be found on GitHub. Here a medium article I wrote about MPC in crypto custody.

Selected Publications

  • A Secure and Scalable Federated Learning Framework without Trusted Aggregators. EMNLP 2021
  • Variance of the Gradient Also Matters: Privacy Leakage from Gradients. IEEE WCCI 2022 (IJCNN 2022)
  • Computational Fuzzy Extractors with Ben Fuller, and Leo Reyzin. 
    Appears in Information and Computation (see Fuzzy Extractor wikipedia)
  • Privacy-Preserving Hierarchical Clustering: Formal Security and Efficient Approximation
    with Dimitrios Papadopoulos, Alina Oprea, Nikos Triandopoulos, ACM CCSW Cloud Computing Workshop, Korea, November 2021
    arXiv version.
  • Top-k Query Processing on Encrypted Databases with Strong Security Guarantees
    with Haohan Zhu, George Kollios, 34th IEEE ICDE, Paris, France, April 2018
    Proceeding version arXiv version Poster.
  • NED: An Inter-Graph Node Metric based on Edit Distance
    with Haohan Zhu, George Kollios, PVLDB 10(6):697-708, 2017 Proceeding version  Poster.
  • GRECS: Approximate Shortest Distance Queries on Encrypted Graphs
    with Seny Kamara, Kobbi Nissim, and George Kollios. 22nd ACM CCS, Denver, Colorado, USA, October 2015
    Proceeding Version, Full version, PDF Slides, Poster
  • Privacy-Preserving Similarity Evaluation of Time Series Data,
    with Haohan Zhu, and George Kollios. 17th EDBT/ICDT, Athens, Greece, March 2014.
    Proceeding Version, Poster, Code
  • Computational Fuzzy Extractors
    with Ben Fuller, and Leo Reyzin. 19th IACR ASIACRYPT, Bangalore, India, December 2013
    Proceeding Version, Full version

About

Previously, I have been working as a leading scientist & tech leader at Amazon Web Service, Inc Facebook, & Apple Inc. from 2016-2021 several startups and spent some time at Microsoft Research in 2015. I received my Ph.D. at Boston University, where I was a member of DBLab and BUSec in the Department of Computer Science at BU. My Ph.D. dissertation is about Privacy-Preserving Queries On Encrypted Databases. See my guest posting on discussing graph privacy: "Graph Encryption: Going Beyond Encrypted Keyword Search", on Seny Kamara's Blog

Journal/Conference Reviewer

IEEE TKDE 2014/2015, IEEE ICDE 2013/2014, SIGMOD 2013/2014
VLDB 2013/2014/2015, EDBT/ICDT 2015/2016
IACR EUROCRYPT 2013, IACR ASIACRYPT 2014...

Courses taught at BU

CS330: Introduction to Algorithms (Taught by Prof. Leonid Levin)
CS235: Algebraic Algorithms (Taught by Prof. Leonid Levin)
CS111: Programming in Java
CS101: Introduction to Computer Science

Past Research

Cryptographic Protocols (HE, MPC, ZKP), Privacy, etc.

In this paper, we proposed GRECS that can support approximate shortest distance queries on the encrypted graph. See the blog entry. (Collaboration with Seny Kamara). We defined a new framework for more secure and scalable federated learning In this we propose an efficient secure top-k queries processing on the encrypted database. I proposed a privacy-preserving protocol for computing the distance between two time-series (see here) and a work that studyies privacy-preserving clustering algorithms. (Collaboration with RSA labs)
Moreover, we defined a new distance metric that is based on the neighborhood topology of two nodes. (see here)