Giuseppe Castiglione
Brighton, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolGiuseppe Castiglione is a machine learning researcher and engineer with 11 years of experience applying representation learning, combinatorial optimization, and AI safety to real-world problems from robotic manipulation to satellite-derived 3D reconstruction. He has led ML teams (Frontier Development Lab), developed physics-informed models and continuous normalizing flows for spatio-temporal forecasting, and shipped safety tooling such as a formal verification library and extensions to the widely used advertorch adversarial robustness package. Comfortable across research and engineering, he has deployed models on embedded and mobile platforms, improved on-device efficiency dramatically, and prototyped RL systems for live trading. Based in Brighton, he combines rigorous academic training (PhD candidacy in Informatics) with hands-on product delivery, and often surfaces unexpected dataset and model biases by probing learned representations.
11 years of coding experience
8 years of employment as a software developer
Doctor of Philosophy - PhD, Informatics, Doctor of Philosophy - PhD, Informatics at University of SussexBachelor of Applied Science (B.A.Sc.), Engineering Physics, Bachelor of Applied Science (B.A.Sc.), Engineering Physics at University of Toronto
English, French
Github Skills (4)
machine-learning10
adversarial-attacks10
python10
pytorch10
Programming languages (2)
Github contributions (5)
BorealisAI/advertorch
Jul 2021 - May 2022
A Toolbox for Adversarial Robustness Research
Role in this project:
ML Engineer
ML EngineerContributions:5 reviews, 26 commits, 6 PRs in 10 months
Contributions summary:Giuseppe primarily contributed to the implementation and refinement of adversarial attack methods within the `advertorch` library. Their work included the addition of new black-box attack strategies like NAttack and improvements to existing methods such as BanditAttack and GenAttack. These changes involved modifications to attack implementations, loss functions, and utility functions.
adversarial-learningpytorchadversarial-attacksbenchmarkingadversarial-perturbations
CaesarQ/advent2021
Dec 2021 - Dec 2021
Contributions:40 pushes, 1 branch in 21 days
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.