Guy Davidson
Machine Learning Researcher at Jane Street
New York, New York, United States
Summary
👤
Senior🎓
Top SchoolGuy Davidson is a machine learning researcher with a PhD in cognitive science and AI from NYU CDS, specializing in cognitively-inspired representations of goals and task structure. With 11 years of hands-on experience spanning industry research roles at Meta FAIR and Jane Street and internships at Microsoft and Amazon, he blends theoretical, empirical, and computational approaches to understanding task representations in large language models. His dissertation produced novel mechanistic and computational advances in how goals are represented and generated, and he has extended that work to interpretability studies of LLMs while at Meta. He has practical ML engineering experience—contributing checkpointing and memory-saving features to the well-known Rainbow deep RL repo—so he moves comfortably between code, experiments, and theory. Based in New York, he balances rigorous research with a pragmatic engineering outlook and a curiosity-driven habit of shipping reproducible experimentation infrastructure. Outside work he’s an avid board gamer and hot-sauce hobbyist, hinting at a playful, experimental mindset that informs his research.
11 years of coding experience
4 years of employment as a software developer
B.Sc. Computational Sciences, B.Sc. Computational Sciences at Minerva UniversityDoctor of Philosophy - PhD Data Science, Doctor of Philosophy - PhD Data Science at New York University
English, Hebrew, Spanish
Github Skills (11)
memory-management10
deep-reinforcement-learning10
pytorch10
deep-learning10
python10
checkpointing9
fileio9
file-access9
checkpoint9
file-handling9
file-processing9
Programming languages (12)
Github contributions (5)
Kaixhin/Rainbow
Sep 2019 - Sep 2019
Rainbow: Combining Improvements in Deep Reinforcement Learning
Role in this project:
ML Engineer
ML EngineerContributions:5 commits, 1 PR, 14 comments in 4 days
Contributions summary:Guy implemented features related to saving and loading the model's memory, crucial for resuming training and preserving the model's state. They added checkpointing capabilities, allowing for saving the model at specified intervals and saving model checkpoints. Additionally, the user made minor quality-of-life improvements and addressed code review comments, contributing to the project's overall maintainability.
deep-reinforcement-learningdeep-learning
guydav/simple-relational-reasoning
Jun 2020 - Aug 2023
Contributions:6 PRs, 237 pushes, 8 branches in 3 years 3 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.