Naoise Holohan
Research Scientist at IBM
Dublin, Dublin 1, Ireland
Summary
🤩
Rockstar🎓
Top SchoolNaoise Holohan is a research scientist with seven years of experience specializing in differential privacy, statistical modelling, and privacy-preserving machine learning, currently leading differential privacy efforts at IBM’s Data Privacy Team. He is the lead developer of IBM’s open-source Differential Privacy Library (diffprivlib) and has contributed practical privacy primitives and variance/bias methods used by practitioners. His background combines a PhD in Applied Mathematics with hands-on work implementing scalable Bayesian and matrix-factorisation solutions for sparse, million-user datasets leveraging multi-core and GPU compute. He has also contributed security-focused tooling to the widely used Adversarial Robustness Toolbox, implementing database reconstruction attacks and reproducible demos. Colleagues describe him as a mathematically rigorous engineer who bridges abstract privacy theory and production-ready software.
7 years of coding experience
4 years of employment as a software developer
PhD, Applied Mathematics, PhD, Applied Mathematics at Trinity College, DublinMSc, Race Car Aerodynamics, Distinction, MSc, Race Car Aerodynamics, Distinction at University of SouthamptonBSc, Mathematics and Applied Mathematics, First-class honours, BSc, Mathematics and Applied Mathematics, First-class honours at University College CorkPhD, Applied Mathematics, PhD, Applied Mathematics at National University of Ireland, Maynooth
Irish, English
Github Skills (11)
data-analysis10
scikit10
statistics10
machine-learning10
adversarial-machine-learning10
adversarial-attacks10
differential-privacy10
python10
scikit-learn10
testing9
jupyter-notebook8
Programming languages (1)
Github contributions (5)
IBM/differential-privacy-library
Jan 2019 - Jan 2023
Diffprivlib: The IBM Differential Privacy Library
Role in this project:
Data Scientist
Data ScientistContributions:14 releases, 22 reviews, 534 commits in 4 years
Contributions summary:Naoise primarily contributed to the `ibm/differential-privacy-library` repository by adding new methods and functionalities related to differential privacy, including adding getVariance methods for Laplace mechanisms and the setEpsilonDelta method. The user's changes involved code modifications, including the introduction of new methods and a .gitignore file. These additions focused on methods related to biases, variance, and general statistical methods of use within the repository, demonstrating the user's commitment to improving the functionality of this library.
pythondata-privacyprivacydifferential-privacymachine-learning
Trusted-AI/adversarial-robustness-toolbox
Nov 2020 - Nov 2020
Adversarial Robustness Toolbox (ART) - Python Library for Machine Learning Security - Evasion, Poisoning, Extraction, Inference - Red and Blue Teams
Role in this project:ML Engineer
ML EngineerContributions:6 commits, 2 PRs, 4 pushes in 7 days
Contributions summary:Naoise focused on implementing and testing database reconstruction attacks within the Adversarial Robustness Toolbox. They developed a `DatabaseReconstruction` class, which is an inference attack, along with a corresponding test suite. The contributions included the implementation of the attack, adding docstrings, and creating a Jupyter notebook to demonstrate its usage. This demonstrates the user's involvement in the development of tools for analyzing and understanding the security of machine learning models.
extractionpythonfairness-mlrobustnessadversarial-machine-learning
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.