Less Wright
AI Software Engineer Cursor
Redmond, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolLess Wright is an AI software engineer based in Redmond with seven years of cross-disciplinary experience building and scaling distributed deep learning systems and production AI. He is a core contributor to PyTorch distributed tooling at Meta and the developer of the widely used Ranger optimizer family (Ranger/Ranger21), demonstrating both research-grade algorithm design and pragmatic engineering. At Cursor he focuses on large-scale training parallelisms and kernel optimizations, and previously led cloud-native computer vision pipelines and a US-patented AI diagnostic architecture at Audere. His background spans systems languages (Rust, Go, C++, C#), blockchain smart-contract work, lab-grown materials robotics, and early Microsoft product engineering—an unusual blend that helps him bridge hardware, chemistry, and ML. A top-rated AI author and consultant with a track record of conference talks and open-source impact, he combines hands-on kernel and optimizer work with product-level deployment experience.
7 years of coding experience
11 years of employment as a software developer
William Howard Taft University (Law School)Certificate - Supervised Machine Learning with scikit-learn, Machine Learning, Completion (no grades given), Certificate - Supervised Machine Learning with scikit-learn, Machine Learning, Completion (no grades given) at Datacamp.comBachelor of Business Administration (BBA), Finance, Bachelor of Business Administration (BBA), Finance at The College of William and MaryCertificate - Python (1 of 5), Python 3, 100%, Certificate - Python (1 of 5), Python 3, 100% at University of MichiganOnline via Coursera - Algorithmic Toolbox: Data Structures and Algorithms., Computer Science, Online via Coursera - Algorithmic Toolbox: Data Structures and Algorithms., Computer Science at University of California, DavisDeep Learning with Python, TensorFlow and Keras 2.0, Artificial Intelligence, Successful Completion (no grades given), Deep Learning with Python, TensorFlow and Keras 2.0, Artificial Intelligence, Successful Completion (no grades given) at DataCamp.comSolid State Chemistry / Materials Science, Solid State Chemistry / Materials Science at MITCourse - Understanding Algorithms for Reinforcement Learning, Reinforcement Learning, None given, Course - Understanding Algorithms for Reinforcement Learning, Reinforcement Learning, None given at Pluralsight.comCertificate - Elements of AI online course, AI, Successful Completion (no grades given), Certificate - Elements of AI online course, AI, Successful Completion (no grades given) at University of HelsinkiAlgorithms, Algorithms at Stanford University
German
Github Skills (9)
pytorch10
machine-learning10
optimizer10
tensor10
deep-learning10
tensorflow10
python10
optimizers10
computer-engineering9
Programming languages (9)
Github contributions (5)
lessw2020/Ranger-Deep-Learning-Optimizer
Aug 2019 - May 2021
Ranger - a synergistic optimizer using RAdam (Rectified Adam), Gradient Centralization and LookAhead in one codebase
Role in this project:
ML Engineer
ML EngineerContributions:41 commits, 5 PRs, 40 pushes in 1 year 8 months
Contributions summary:Less primarily contributed to the development and optimization of a deep learning optimizer, Ranger, within the repository. They implemented the core logic for the optimizer, including RAdam and Lookahead functionalities, and later integrated gradient centralization. The contributions include parameter adjustments, code cleanup, and the addition of new features. The user's work also involved performance improvements and fixing potential save/load issues.
radamlookaheadoptimizerrangeradam
pytorch-labs/applied-ai
Mar 2024 - Mar 2025
Applied AI experiments and examples for PyTorch
Contributions:17 reviews, 34 PRs, 148 pushes in 11 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.