Hadi Salman
Research Lead at OpenAI
San Francisco, California, United States
Summary
👤
Senior🎓
Top SchoolHadi Salman is a research lead at OpenAI with nine years of experience at the intersection of machine learning, robotics, and systems, combining roles at MIT, Microsoft Research, Uber ATG, CMU and AUB. He holds a PhD in Computer Science from MIT and an MS in Robotics from CMU, and his work ranges from adversarial robustness research — contributing to MadryLab’s widely used robustness library — to applied research for production-scale ML systems. Hadi bridges deep academic rigor with practical engineering, shipping bug fixes, checkpointing improvements, and model compatibility changes that smooth the path from experimentation to deployment. Based in San Francisco, he thrives on translating cutting-edge research into reliable, deployable systems and often works on less-visible but critical engineering details that keep complex ML pipelines running.
9 years of coding experience
8 years of employment as a software developer
Doctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at Massachusetts Institute of TechnologyMS in Robotics School of Computer Science, MS in Robotics School of Computer Science at Carnegie Mellon UniversityBS in Mathematics - BE in Mechanical Engineering, BS in Mathematics - BE in Mechanical Engineering at American University of Beirut
Github Skills (13)
neural-network10
robust10
pytorch10
machine-learning10
trainings10
python10
modeling10
robustness10
ml-deployment9
deep-learning9
continuous-deployment9
imagenet8
tensorflow3
Programming languages (4)
Github contributions (5)
MadryLab/robustness
Apr 2020 - Feb 2022
A library for experimenting with, training and evaluating neural networks, with a focus on adversarial robustness.
Role in this project:
ML Engineer
ML EngineerContributions:25 commits, 17 PRs, 24 pushes in 1 year 9 months
Contributions summary:Hadi made several contributions focused on improving the functionality and maintainability of the robustness library, which specializes in adversarial robustness for neural networks. They added support for normal inference within the AttackerModel, fixed a bug related to loading subsets of datasets, and refactored code by moving custom modules and fixing CIFAR densenet models. They also introduced changes related to checkpoint management and fixed backwards compatibility for saved models, indicating involvement in the training and model deployment aspects of the library.
pytorchexperimentingrobustnessdeep-learningadversarial
Hadisalman/gym-gazebo-hadi
Nov 2017 - Jun 2018
Contributions:33 commits, 12 pushes, 5 branches in 6 months
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.