Nevan Wichers
Anthropic Fellow
Greater Seattle Area United States
Summary
🤩
Rockstar🎓
Top SchoolNevan Wichers is a senior software engineer and ML researcher with 11 years of experience focusing on AI safety and trustworthy models, currently an Astra Research Fellow after a decade at Google. He develops practical safety techniques—most notably Gradient-Based Red Teaming to automatically find prompts that elicit unsafe LLM outputs—and has contributed to video prediction, referring-expression resolution, and imitation learning toolkits in open-source repos. His work has influenced industry practice (e.g., inoculation prompting adopted at Anthropic) and improved model robustness and metrics across multiple projects. A seasoned mentor and educator, he teaches ML courses at Google and reviews for top conferences, bridging hands-on engineering, publishable research, and deployment constraints on real devices. Based in Seattle, he combines deep algorithmic expertise with a pragmatic focus on making powerful models safer in the wild.
11 years of coding experience
10 years of employment as a software developer
Edmonds CollegeComputer Engineering, Computer Engineering at Seattle University
Github Skills (10)
imitation-learning10
machine-learning10
tensorflow10
python10
pytorch9
algorithms8
data-structures8
algorithm8
data-structure8
gymnasium7
Programming languages (6)
Github contributions (5)
HumanCompatibleAI/imitation
Mar 2019 - Mar 2019
Clean PyTorch implementations of imitation and reward learning algorithms
Role in this project:
ML Engineer
ML EngineerContributions:34 commits, 5 PRs, 24 pushes in 12 days
Contributions summary:Nevan contributed to the implementation of imitation and reward learning algorithms, specifically focusing on building the core components for adversarial imitation learning (AIRL) and generative adversarial imitation learning (GAIL). This includes creating abstract interfaces for discriminators and reward networks, defining loss functions, and structuring the training procedure for both AIRL and GAIL. The user's commits demonstrate an understanding of PyTorch, TensorFlow, and core concepts in inverse reinforcement learning.
pytorchimplementationsreinforcement-learningcleanmachine-learning
brain-research/long-term-video-prediction-without-supervision
Jan 2018 - Nov 2021
Implementation of Hierarchical Long-term Video Prediction without Supervision
Contributions:21 commits, 2 PRs, 14 pushes in 3 years 10 months
pytorchlong-termvideo-predictionpredictiondeep-learning
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.