Apostolos Modas
Senior AI Research Engineer at SonyAI
Switzerland
Summary
👤
Senior🎓
Top SchoolApostolos Modas is a Senior AI Research Engineer based in Zürich with nine years of experience at the intersection of deep learning, computer vision, and trustworthy generative AI. He holds a PhD from EPFL where he developed geometric frameworks linking adversarial robustness to generalization and advanced methods for distribution-shift resilience, work published in venues including Nature, NeurIPS and CVPR. At Sony AI he leads privacy-preserving, ethics-first projects that deploy foundation models for large-scale PII anonymization and curate diverse image datasets—overseeing the annotation of 35K+ human-centric images to improve fairness evaluations. An active open-source contributor, he’s implemented and refined core evasion attacks in IBM’s Adversarial Robustness Toolbox, blending theoretical rigor with practical security engineering. Colleagues know him for translating cutting-edge research into production-ready systems that prioritize transparency, regulatory compliance, and measurable bias mitigation.
9 years of coding experience
4 years of employment as a software developer
Master’s Degree Signal and Information Processing and Learning, Master’s Degree Signal and Information Processing and Learning at Ethnikon kai Kapodistriakon Panepistimion AthinonDoctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at EPFL
English, German, Greek
Github Skills (8)
machine-learning10
eval10
adversarial-attacks10
python10
batch-processing9
numpy9
tensorflow9
pytorch9
Programming languages (2)
Github contributions (5)
Trusted-AI/adversarial-robustness-toolbox
Mar 2021 - Mar 2021
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:9 commits, 2 PRs, 4 comments in 8 days
Contributions summary:Apostolos's primary contribution involves implementing and refining adversarial attack algorithms within the Adversarial Robustness Toolbox (ART) library. They focused on improving the `BoundaryAttack` and `HopSkipJump` evasion attacks by addressing criteria for targeted and untargeted attacks, as well as ensuring the return of adversarial examples with minimal L2 norms. Their work also included refactoring the code to align with PEP8 style guidelines and incorporating a `batch_size` parameter to improve performance, ultimately demonstrating their expertise in machine learning security and evasion techniques.
extractionpythonfairness-mlrobustnessadversarial-machine-learning
amodas/PRIME-augmentations
Dec 2021 - Dec 2022
Contributions:6 commits, 2 pushes, 1 comment in 1 year
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.