Edoardo Debenedetti
Doctoral Researcher at Department of Computer Science (D-INFK), ETH Zürich
Zurich, Zurich, Switzerland
Summary
👤
Senior🎓
Top SchoolEdoardo Debenedetti is a PhD researcher at ETH Zürich specializing in the security and privacy of machine learning, with a particular focus on AI agents and prompt-injection defenses. His work is supported by a CYD Doctoral Fellowship and builds on a strong research foundation from EPFL and collaborations with Princeton and Google DeepMind. He has practical experience shipping security-focused tooling and defenses—contributing backend and DevOps improvements to RobustBench and stateful evaluation features to the widely used AutoAttack implementation. Edoardo pairs academic rigor with industry practice through internships at Google, Meta, Bloomberg and the Swiss Cyber-Defence Campus, and brings a systems-minded engineering background from earlier software roles. Fluent in cross-cultural research settings after a PoliTong exchange in Shanghai and an unusual Military Naval College humanities background, he blends disciplined problem-solving with a security-first approach to ML.
13 years of coding experience
1 year of employment as a software developer
Exchange Student Information Technology Engineering, Exchange Student Information Technology Engineering at Tongji UniversityHumanities High School Diploma Humanities, Humanities High School Diploma Humanities at Scuola Navale Militare Francesco MorosiniBachelor of Science - BS Computer Engineering, Bachelor of Science - BS Computer Engineering at Politecnico di TorinoMaster of Science - MS Computer Science, Master of Science - MS Computer Science at EPFLDoctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at ETH Zürich
English, French, Italian
Stackoverflow
95reputation
2kreached
1answer
3questions
Github Skills (23)
json10
pytorch10
benchmark10
adversarial-attacks10
python10
machine-learning10
benchmarking10
zoo10
robustness10
robust10
jinja29
data-structures8
file-handling8
data-structure8
devops7
Programming languages (7)
Github contributions (5)
RobustBench/robustbench
Jan 2021 - Dec 2022
RobustBench: a standardized adversarial robustness benchmark [NeurIPS 2021 Benchmarks and Datasets Track]
Role in this project:
Backend Developer & DevOps Engineer
Backend Developer & DevOps EngineerContributions:4 releases, 17 reviews, 210 commits in 1 year 11 months
Contributions summary:Edoardo contributed to the project by creating a leaderboard generator, which processes JSON results and renders an HTML table for display. The changes involved modifying Python scripts and Jinja2 templates. Additionally, the user updated the signature of the `list_available_models` function and modified several files related to model definitions within the model zoo. This indicates work across both backend utilities and potentially model management aspects of the project.
pytorchbenchmarkingrobustnessadversarial-machine-learningdeep-learning
fra31/auto-attack
Nov 2022 - Dec 2022
Code relative to "Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attacks"
Role in this project:ML Engineer
ML EngineerContributions:6 commits, 5 comments in 21 days
Contributions summary:Edoardo primarily focused on implementing and refining the `AutoAttack` class and related state management within the repository. Their contributions involved adding a state implementation for robust evaluation of adversarial attacks. The user added functionality to save and restore evaluation states, including robust flags and attack run status, to improve the efficiency of the attacks. The user also added a warning for saved examples when a state path is provided.
adversarial-attacksdiverserelativerobustnessadversarial-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.