Cassidy Laidlaw is a PhD student in computer science at UC Berkeley and an experienced engineer with 11 years building production software, research systems, and ML tools across academia, government, and startups. They blend hands-on consulting and product development—ranging from rebuilding a payments-enabled mentoring platform to deploying security-focused analytics for DHS—with rigorous research into adversarial machine learning. As an open-source contributor to the CleverHans project, Cassidy implemented and hardened the SPSA adversarial attack in PyTorch, adding L2 support, clamping, and sanity checks that improve benchmarking reliability. Comfortable translating between research and production, they have led end-to-end projects in web, mobile, and data science while presenting technical work to stakeholders including federal agencies. Based in the San Francisco Bay Area, Cassidy brings a rare mix of applied ML, systems engineering, and academic depth that helps organizations move cutting-edge models into reliable, real-world use.
11 years of coding experience
3 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of California, Berkeley
Bachelor’s Degree, Mathematics and Computer Science, Minor in Middle East Studies, 4.0 GPA, Bachelor’s Degree, Mathematics and Computer Science, Minor in Middle East Studies, 4.0 GPA at University of Maryland
An adversarial example library for constructing attacks, building defenses, and benchmarking both
Role in this project:
ML Engineer
Contributions:8 commits, 2 PRs, 8 comments in 6 months
Contributions summary:Cassidy implemented and refined the SPSA (Simultaneous Perturbation Stochastic Approximation) attack within the PyTorch framework, a core component of the CleverHans library. They focused on adding features like L2 norm support and clamping, improving the attack's robustness and adherence to defined constraints. The user also addressed potential issues by including additional sanity checks, ensuring code reliability. Overall, the contributions centered around expanding and improving the adversarial attack capabilities of the library.
An open source framework that provides a simple, universal API for building distributed applications. Ray is packaged with RLlib, a scalable reinforcement learning library, and Tune, a scalable hyperparameter tuning library.
Contributions:29 pushes, 5 branches in 2 years 1 month
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.